JP2016206879A - Parameter classification device, and movable body simulated motion generating device and movable body motion generating device based on parameter classification device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a parameter classification device capable of setting parameters that may generate simulated motion of a movable body conforming to prescribed evaluation criteria easily and efficiently, and a movable body motion generating device based on the parameter classification device.SOLUTION: On the basis of simulated motion SM (vector ds, vector es) generated in association with each of the plurality of sample parameter sets vector ds, an evaluation determination part 112 determines whether a classification target parameter sample value included in each of the plurality of sample parameter sets vector ds is a value allowing generation of simulated motion conforming to prescribed evaluation criteria. In accordance with the determination result from the evaluation determination part 112, a selection permissible region determination part 113 classifies the whole set of classification target parameter sets vector d into a first region R1 and a second region R2, and determines the first region R1 or a partial region thereof as a selection permissible region of the classification target parameter sets vector d.SELECTED DRAWING: Figure 4

Description

  The present invention relates to a parameter classification device, a moving body simulated motion generation device using the parameter classification device, and a mobile body motion generation device.

  Various proposals have been made on vehicle motion models representing vehicle motion and their applications.

  For example, a driver model describing a probability distribution in which each data in an N-dimensional space exists is obtained by using time-series data of N types of feature amounts detected as a driver drives a vehicle and learning data as an EM algorithm (expectation). There has been proposed an apparatus that estimates driving behavior close to the driving characteristics of the driver by using a GMM (Gaussian mixture model) calculated by a value maximization method (see, for example, Patent Document 1). Based on the driving behavior by the driver model, the vehicle motion of the driver can be reproduced.

  In addition, by determining parameters of an optimal velocity model representing the motion of the vehicle based on observation data of the motion of the individual vehicle traveling on the actual road, the motion of the individual vehicle is reproduced. A method has been proposed (see, for example, Non-Patent Document 1).

Japanese Patent No. 4781104

Masako Bando and five others, “Phenomenological Study of Dynamic Model of Traffic Flow”, [online] Internet <http: // cphys. meijo-u. ac. jp / pdf / jpif5. pdf>

  In recent years, research on traffic simulation has been active. Such a simulation of traffic conditions can also be used effectively to determine the motion of a moving object in an automatic driving algorithm.

  When determining the motion of a vehicle, it is preferable to simulate a wide variety of traffic conditions, but in order to simulate a wide variety of traffic conditions, it is possible to generate a wide variety of simulated motions of a moving object. is necessary.

  However, in the techniques described in Patent Document 1 and Non-Patent Document 1, in order to generate a wide variety of simulated motions of a moving body, it is possible to generate a wide variety of driver models or determine various parameters of a vehicle motion model. For this reason, a huge amount of observation data is required.

  On the other hand, when a driver model is generated or a vehicle motion model parameter is determined by a random number or the like, for example, the driver model or the vehicle motion model may generate a simulated motion that deviates from the traffic situation.

  In view of these problems, the present invention is a parameter classification device capable of setting a parameter capable of generating a simulated motion of a moving body that conforms to a predetermined evaluation criterion, and a moving body simulated motion generation using the parameter classification device. An object of the present invention is to provide a device and a moving body motion generation device.

The parameter classification apparatus of the present invention
A first motion that generates a simulated motion of the moving body using a moving body motion model that is described by a parameter set that includes a plurality of parameters including a predetermined type of classification target parameter and that represents the motion of the moving body. A moving body simulated motion generation unit;
The first moving body simulated motion generation unit is caused to generate a simulated motion of the moving body using each of a plurality of sample parameter sets that are a plurality of samples of the parameter set, and the plurality of sample parameter sets Based on the simulated motion generated corresponding to each of the plurality of sample parameter sets, the classification target parameter sample value that is the value of the classification target parameter included in each of the plurality of sample parameter sets is a simulated motion that conforms to a predetermined evaluation criterion. An evaluation determination unit that determines whether or not the value can be generated;
In accordance with the determination result of the evaluation determination unit, the entire set of classification target parameters, a first region including the classification target parameters capable of generating a simulated motion that conforms to the evaluation criterion, and a simulation that conforms to the evaluation criterion Classifying the movement into the second region including the classification target parameter that cannot be generated;
A selection permissible area determining unit that determines the first area or a partial area thereof as a selection permissible area of the classification target parameter for generating a simulated motion of the moving body is provided.

  Supplementally, in the present invention, the classification target parameter included in the parameter set includes one or more parameters.

  The plurality of sample parameter sets are parameter sets having different values for at least one parameter included in each of the sample parameter sets. In addition, all or some of the plurality of sample parameter sets are parameter sets in which values of classification target parameters included in the sample parameter sets are different from each other.

  Here, the value of the classification target parameter included in each parameter set means the value of the single classification target parameter itself when the classification target parameter included in the parameter set consists of one parameter. In the case where the classification target parameter included in is composed of a plurality of parameters, it means a set of the respective values of the plurality of classification target parameters.

  Therefore, when there are a plurality of classification target parameters included in each parameter set, the classification target parameters included in each of any two parameter sets are different from each other, indicating that at least one classification target included in each parameter set is included. This means that the parameter values are different from each other.

  In the present invention, the first region is a region in which all or most of the classification target parameters included in the region are classification target parameters capable of generating a simulated motion that meets the evaluation criteria. On the other hand, the second region is a region in which all or most of the classification target parameters included in the region are classification target parameters that cannot generate a simulated motion that meets the evaluation criteria.

  In other words, the content rate in the first region of the classification target parameter capable of generating a simulated motion that meets the evaluation criterion is higher than the content rate in the second region.

  According to the parameter classification device of the present invention, according to the determination result of the evaluation determination unit, the entire set of the classification target parameters includes the classification target parameter capable of generating a simulated motion that meets the evaluation criterion. And a second region including the classification target parameter that cannot generate a simulated motion that conforms to the evaluation criterion, and the first region or its partial region is a classification target parameter for generating a simulated motion of a moving object Is determined as a selection allowable region.

  By using the classification target parameter included in the selection allowable region for the generation of the simulated motion of the moving body, the probability that the generated simulated motion of the moving body meets the evaluation criterion is increased.

  Further, the entire set of classification target parameters is classified into the first area and the second area based on the sample values of the classification target parameters included in each of the plurality of sample parameter sets, which is a subset of the total set of classification target parameters. Therefore, selection of a classification target parameter that can generate a simulated motion that conforms to a predetermined evaluation criterion is easily and efficiently performed.

In the parameter classification device of the present invention,
The parameter set preferably includes a moving body motion characteristic parameter indicating a motion characteristic of the moving body and a surrounding object motion characteristic parameter indicating a motion characteristic of an object existing around the moving body.

  According to the parameter classification device of the configuration, the parameter set includes the moving body motion characteristic parameter indicating the motion characteristic of the moving body and the surrounding object motion characteristic parameter indicating the motion characteristic of the object around the moving body. The simulated motion of the moving body is generated in a manner that appropriately reflects the motion characteristics of the object and the motion characteristics of the objects around the moving body.

In the parameter classification device of the configuration,
The parameter set includes the classification target parameter and a classification non-target parameter that is a non-classification parameter,
The evaluation determination unit
The plurality of sample parameter sets are a plurality of samples in which the classification target parameter sample values are the same and the classification non-target parameter sample values that are values of the classification non-target parameters included in the sample parameter set are different from each other. Including parameter sets,
Of the total number of the simulated exercises generated corresponding to each of the plurality of sample parameter sets including the same classification target parameter sample values, the ratio of the number of simulated exercises satisfying a predetermined condition is equal to or greater than a predetermined ratio. It is preferable to determine that the same classification target parameter sample value is a value that can generate a simulated motion that conforms to a predetermined evaluation criterion.

  In the present invention, the value of the classification non-target parameter has the same meaning as the value of the classification target parameter.

  According to the parameter classification device having the above configuration, the number of simulated exercises satisfying a predetermined condition out of the total number of simulated exercises generated corresponding to each of a plurality of sample parameter sets including classification target parameters having the same value. The ratio being equal to or greater than a predetermined ratio is a necessary condition for determining that the same classification target parameter sample value is a value capable of generating a simulated motion that conforms to a predetermined evaluation criterion.

  As a result, in the first region, it is determined that a simulated motion that satisfies a predetermined condition can be generated with a probability greater than or equal to a predetermined probability as a classification target parameter included in the first region capable of generating a simulated motion that meets a predetermined evaluation criterion. The classified classification target parameters are classified.

  Then, this first area or its partial area is determined as a selection allowable area for the classification target parameter.

  By using the classification target parameter included in the selection allowable region for generating the simulated motion of the moving object, the probability that the simulated motion of the moving object that satisfies the predetermined condition is generated with a probability of a predetermined value or higher is increased.

In the parameter classification device of the configuration,
It is preferable that the predetermined condition is a condition that the moving body does not contact another object.

  According to the parameter classification device having the above configuration, the probability that a simulated motion of a moving body that does not contact other objects with a predetermined probability or higher is generated.

In these parameter classifiers,
The classification target parameter is preferably the moving body motion characteristic parameter.

  According to the parameter classification device having the above configuration, it is possible to set a moving body motion characteristic parameter capable of generating a simulated motion that conforms to a predetermined evaluation criterion in various traveling environments generated by various surrounding object motion characteristic parameters. I can do it.

In the parameter classification device of the present invention,
The evaluation determination unit includes a classification target parameter criterion which is a value of a classification target parameter serving as a reference and a simulated movement of the mobile unit generated by using the classification target parameter sample value in the first moving target simulated movement generation unit. If the similarity with the simulated motion of the mobile body generated by the first mobile body simulated motion generation unit using a value is equal to or higher than a predetermined similarity, the classification target parameter sample value is a predetermined value It is preferable to determine that the value is a value that can generate a simulated motion that meets the evaluation criteria.

  According to the parameter classification device of the configuration, the classification target parameter that can generate the simulated motion of the moving object whose similarity to the simulated motion of the moving object generated using the classification target parameter reference value is equal to or higher than the predetermined similarity Can be set.

  In the parameter classification apparatus having the configuration, it is preferable that the classification target parameter reference value is set based on observation data obtained by observing the movement of a moving object in an actual environment.

  According to the parameter classification device having the above configuration, it is possible to set a classification target parameter that can generate a simulated motion similar to the motion of a moving object in a real environment.

In these parameter classifiers,
The classification target parameter reference value is preferably a plurality of different classification target parameter reference values.

  According to the parameter classification device having the configuration, diversity of classification target parameter reference values can be ensured.

In these parameter classifiers,
The selection allowable region determining unit
Based on the observation data obtained by observing the movement of the moving object in the real environment, the existence area of the classification target parameter in the entire set of classification target parameters,
It is preferable that an overlapping area between the classification target parameter existing area and the first area is determined as the selection allowable area.

  According to the parameter classification device of the configuration, by generating the simulated motion of the moving object using the classification target parameter included in the selection allowable region, the simulated motion of the moving object deviated from the motion of the moving object in the real environment. Generation can be avoided.

In the parameter classification device of the present invention,
The classification that has a relatively large influence on the conformity determination of the simulated motion in which the change in the value of the classification target parameter is generated based on the definition area for each value of the classification target parameter in the selection allowable area It is preferable that a target classification target parameter determination unit that determines the target parameter as a target classification target parameter is provided.

  According to the parameter classification apparatus having the configuration, a classification target parameter that has a relatively large influence on the determination of conformity to the evaluation standard of the simulated motion in which a change in value is generated is determined as the target classification target parameter. By generating the sample parameter set around the target classification target parameter, the classification target parameter sample can be easily and efficiently created.

The moving body simulated motion generation device of the present invention is
A second moving body simulated motion generation unit configured to generate a simulated motion of the moving body using the classification target parameter included in the selection allowable region determined by the parameter classification device of the present invention and the moving body motion model; It is characterized by providing.

  According to the moving body simulated motion generation apparatus, a simulated motion of the moving body is generated using the classification target parameter and the moving body motion model included in the selection allowable region. For this reason, the probability that the simulated motion of the moving body conforms to the predetermined evaluation standard is relatively high.

The mobile body motion generation device of the present invention includes:
An external environment recognition unit for recognizing information on the external environment as a value of the surrounding object motion characteristic parameter;
Using the moving body motion characteristic parameter included in the selection allowable region determined by the parameter classification device of the present invention, the moving body motion model, and information about the external environment recognized by the external environment recognition unit, A moving body motion determining unit that determines the motion of the moving body of
And an operation control unit that controls the operation of the device of the target mobile body so as to realize the motion of the target mobile body determined by the mobile body motion determination unit.

  According to the mobile body motion generation device configured as described above, an arbitrary mobile body motion characteristic parameter included in the selection allowable region generated by the parameter classification device of the present invention is selected as the mobile body motion characteristic parameter of the target mobile body. The Here, the selection allowable region is a region including a moving body motion characteristic parameter capable of generating a simulated motion that meets a predetermined evaluation criterion or a partial region thereof. Therefore, by determining the motion of the target mobile body using the mobile body motion characteristic parameter included in the selection allowable region, the probability that the determined motion of the target mobile body meets the predetermined evaluation criteria is relatively high. To be high.

  That is, according to the mobile body motion generation device having the above configuration, it is possible to easily and efficiently generate the motion of the target mobile body having a relatively high probability of meeting a predetermined evaluation criterion.

  Note that “the selection allowable region determined by the parameter classification device of the present invention” is a selection allowable region determined by using the moving body motion characteristic parameter as a classification target parameter. The selection allowable region may be determined by any of the parameter classification apparatuses.

The figure which shows the whole structure outline | summary of the parameter classification | category apparatus and moving body simulated exercise | movement production | generation apparatus of 1st Embodiment. FIG. 2A is a diagram for explaining the process of defining the position of each mobile body, FIG. 2B is a diagram for explaining the position of the gap, and FIG. 2C is a lane change diagram. FIG. 2D is a diagram for explaining the position of the gap to be considered, and FIG. 2D is a diagram for explaining the position of the gap when lane change is determined. The flowchart of a selection permissible area | region determination process. FIG. 4A is a diagram illustrating a virtual distribution in a two-dimensional parameter space of mobile body motion characteristic parameters, FIG. 4A is a diagram illustrating a relationship between a mobile body motion characteristic parameter reference value and a mobile body motion characteristic sample parameter set, and FIG. FIG. 4C is a diagram showing the outer edge OE of the selection allowable region, FIG. 4C is a diagram showing the distribution of the mobile body motion characteristic sample parameter set with valid labels and the mobile body motion characteristic sample parameter set with invalid labels, and FIG. 4D is the first teacher signal FIG. 4E is a diagram showing the distribution of the mobile body motion characteristic sample parameter set to which the second motion signal is attached, and FIG. 4E shows the entire set of mobile body motion characteristic parameters in the first region. The figure which shows the range of the 1st area | region and 2nd area | region when it divides | segments into 2nd area | region. The flowchart of a mobile body simulated exercise | movement production | generation process. A graph showing simulated motion of a moving body created from various moving body motion characteristic parameters with time as a horizontal axis, and each movement when the simulated motion is generated using the moving body motion characteristic parameters described in Non-Patent Document 1. A graph showing the position of the body. This graph shows simulated motion of a moving body created from various moving body motion characteristic parameters with time as the horizontal axis, and is simulated using the moving body motion characteristic parameters generated by the moving body simulated motion generation process of this embodiment. The graph which shows the position of each moving body at the time of producing | generating a motion. 6B is a graph showing simulated motion of a moving body created from various moving body motion characteristic parameters with time as a horizontal axis, and the simulated motion is generated using a moving body motion characteristic parameter different from the moving body motion characteristic parameter of FIG. 6B. The graph which shows the position of each moving body in case. A graph showing simulated motion of a moving body created from various moving body motion characteristic parameters with time as a horizontal axis, and each movement when the simulated motion is generated using the moving body motion characteristic parameters described in Non-Patent Document 1. A graph showing the speed of the body. This graph shows simulated motion of a moving body created from various moving body motion characteristic parameters with time as the horizontal axis, and is simulated using the moving body motion characteristic parameters generated by the moving body simulated motion generation process of this embodiment. The graph which shows the position of each moving body at the time of producing | generating a motion. 6B is a graph showing simulated motion of a moving body created from various moving body motion characteristic parameters with time as a horizontal axis, and the simulated motion is generated using a moving body motion characteristic parameter different from the mobile body motion characteristic parameter of FIG. 6E. The graph which shows the position of each moving body in case. The distance of each inter-vehicle distance when generating a simulated motion for changing lanes using the moving body motion characteristic parameter generated by the moving body simulated motion generation process of the present embodiment, with the inter-vehicle distance as the horizontal axis and the appearance frequency as the vertical axis. 7A is a graph comparing the appearance frequency and the appearance frequency of each inter-vehicle distance derived from the actual environment measurement data, FIG. 7A is a graph showing the inter-vehicle distance from the preceding vehicle before the lane change, and FIG. 7B is the pre-lane change FIG. 7C is a graph showing the inter-vehicle distance with the preceding vehicle after the lane change, and FIG. 7D is a graph showing the inter-vehicle distance with the following vehicle after the lane change. The figure which shows the whole structure outline | summary of the parameter classification | category apparatus and moving body simulated exercise | movement production | generation apparatus of 2nd Embodiment. The flowchart of an attention classification object parameter determination process. The figure which shows the whole structure outline | summary of the parameter classification | category apparatus and moving body motion production | generation apparatus of 3rd Embodiment. Flow chart of moving body operation control processing

  Hereinafter, the first embodiment of the present invention will be described with reference to FIGS. 1 to 7 by taking the parameter classification device 100 and the moving body simulated motion generation device 200 as examples.

(Configuration of parameter classification device)
As shown in FIG. 1, the parameter classification device 100 includes a first arithmetic processing unit 110 and a first storage unit 120.

  The parameter classification device 100 is configured by one or a plurality of computers including an arithmetic processing device and a storage device.

  The arithmetic processing unit of one or more computers reads and executes the parameter classification program stored in the storage device, thereby configuring the first arithmetic processing unit 110.

  The first arithmetic processing unit 110 includes a first moving body simulated exercise generation unit 111, an evaluation determination unit 112, and a selection allowable region determination unit 113.

  The first moving body simulated exercise generation unit 111, the evaluation determination unit 112, and the selection allowable region determination unit 113 are configured to execute a parameter selection allowable region determination process described later.

  Here, the first moving body simulated motion generation unit 111 uses each moving body using a moving body motion model for each moving body in a virtual space where the plurality of moving bodies travel in one or a plurality of lanes. Is configured to generate a simulated motion.

  In the present embodiment, the moving body is a vehicle (four-wheel automobile, two-wheel automobile, bicycle, etc.) or a pedestrian traveling on the road. Instead, the first moving body simulated motion generation unit 111 may generate a simulated motion of a moving body other than a vehicle such as an aircraft or a ship or a pedestrian.

  The moving body motion model is a model that is described by a parameter set and expresses the motion of one moving body. In this embodiment, the optimal speed and acceleration of the moving body when the lane is not changed is determined. A speed model and a lane change model that determines whether or not a lane change is necessary are used.

  The parameter set includes a set of moving body motion characteristic parameters (corresponding to the “classification target parameter” of the present invention) indicating the motion characteristics of the moving body (hereinafter referred to as “moving body motion characteristic parameter set”), and A set of surrounding object motion characteristic parameters (corresponding to the “classification non-target parameter” of the present invention) indicating the motion characteristics of an object (including one or both of another moving body and a stationary object) existing around the moving body. (Hereinafter referred to as “ambient object motion characteristic parameter set”).

  Details of the moving body motion model and parameters will be described later.

  The first storage unit 120 includes storage devices such as a main storage device, a secondary storage device, and an auxiliary storage device. Instead, the first storage unit 120 may be configured by a database server provided by a cloud service.

  The first storage unit 120 includes pre-stored initial setting information 121, pre-stored real environment observation data 122, pre-stored mobile body motion characteristic parameter reference value information 123, and parameter selection allowable region determination described later. The selection permissible area information 124 generated in the processing is stored.

  The moving body simulated exercise generating apparatus 200 includes a second arithmetic processing unit 210 and a second storage unit 220.

  The moving body simulated motion generation device 200 is configured by one or a plurality of computers including an arithmetic processing device and a storage device.

  The arithmetic processing unit of one or more computers reads and executes the parameter classification program stored in the storage device, thereby configuring the second arithmetic processing unit 210.

  The second arithmetic processing unit 210 includes a second moving body simulated exercise generation unit 211.

  The second storage unit 220 includes a storage device such as a main storage device, a secondary storage device, and an auxiliary storage device. Instead, the second storage unit 220 may be configured by a database server provided by a cloud service.

  The parameter classification device 100 and the moving body simulated motion generation device 200 may be configured by physically separate hardware (computers), respectively, or a part (for example, a storage device) may be shared and the rest (for example, an arithmetic processing device). May be configured by separate hardware, or all may be configured by the same hardware.

  When the first storage unit 120 of the parameter classification device 100 and the second storage unit 220 of the moving body simulated exercise generating apparatus 200 are configured with separate hardware, the moving body simulated exercise generating apparatus 200 uses a network such as the Internet. Via the first storage unit 120 of the parameter classification device 100, the selection allowable region information 124 is acquired and stored in the second storage unit 220.

(Moving body motion model, parameter set, real environment observation data and moving body motion characteristic parameter reference value set)
Here, the moving body motion model, the parameter set, the real environment observation data, and the moving body motion characteristic parameter reference value set used in this embodiment will be described.

  The moving body motion model of the present embodiment is based on the para-metaset at time t in a traffic situation where N (N ≧ 1) moving bodies n (n = 1,..., N) exist. Is a model for determining the simulated motion (position ↑ x (n, t + Δt)) of the moving body n in FIG.

  In this specification, “↑” is used as a symbol representing a vector (vertical vector).

  Here, the meaning of each parameter is as follows.

  As shown in FIG. 2A, the position of each mobile object n (n = 1,..., N) is based on one point of one mobile object n (or a reference mobile object) at each time t. Are expressed with the traveling direction of the moving body n as the first axis and the direction perpendicular to the traveling direction as the second axis. Hereinafter, the positions of the moving body n at the time t on the first axis and the second axis are collectively expressed as a vector ↑ x (n, t).

  In addition, the position ↑ x (n, t) of each moving body n at time t is collectively expressed as a matrix X (t).

The mobile body motion characteristic parameter set ↑ d (n) is expressed by the following equation (3) using each mobile body motion characteristic parameter d _i (n) (i = 1, 2,... 10). .

Here, the moving body motion characteristic parameter d _i (n) (i = 1,..., 5) is the movement used in the following equations (4) to (5) representing the optimum speed model constituting the moving body motion model. It is a body movement characteristic parameter.

The optimal speed model includes the moving body motion characteristic parameter d _i (n), the position ↑ x (n, t) of the moving body n and its time partial differential value, and the preceding running of the moving body n and the moving body n. The speed V (x (n, t + Δt)) and acceleration A (x (n, t + Δt) as a simulated motion at time t + Δt of the moving body n using the position ↑ x _n ^{n + 1} (t) of the inter-vehicle distance with the car n + 1. )).

here,

Here, supplementing each moving body motion characteristic parameter d _i (n) (i = 1,..., 5), d ₁ (n) is the maximum acceleration of the moving body n, and d ₂ (n) is the moving body. n speed hysteresis, d ₃ (n) is the maximum speed of the moving body n, d ₄ (n) is the degree of tracking of the moving body n with respect to the moving body n + 1 ahead of the moving body n, and d ₅ (n) is , Means the minimum target inter-vehicle distance of the moving body n.

Further, d _i (n) (i = 6,..., 10) is a parameter used in the following equations (6) to (9) representing the lane change model constituting the moving body motion model.

  Note that the lane change model constituting the moving body motion model is a model that represents conditions for changing the lane of the moving body n at time t + Δt based on the parameter set at time t.

  However,

  Here, as shown in FIG. 2B, the gap is a space in which no other moving body n exists among the spaces formed on the lane adjacent to the lane in which the moving body n is traveling at time t. (Space surrounded by a broken line indicated by gap1, gap2, gap3, gap4, and gap5 in FIG. 2B).

  Further supplementally, in determining whether to change the lane to the right lane by the moving body n at the time t, the first moving body simulated movement generation unit 111, as shown in FIG. It is determined whether or not the gap s (s = gap1 to gap3) existing in the right lane calculated from the position satisfies all the equations (6) to (9) of the lane change model.

  When a gap s that satisfies all of the equations (6) to (9) of the lane change model is found at time t (s = gap2 in FIG. 2D), the first moving body simulated motion generation unit 111 moves As a simulated motion of the body n at time t, a simulated motion of moving from the current position of the moving body n to the gap s = gap2 is generated.

  If the gap s satisfying all the equations (6) to (9) of the lane change model is not found at time t, the first moving body simulated motion generation unit 111 simulates the moving body n at time t. As the motion, a simulated motion is generated in which the moving body n continues to travel in the lane in which the moving body n is currently traveling.

Here, supplementing each moving body motion characteristic parameter d _i (n) (i = 6,..., 10), d ₆ (n) is the ratio of the speed at which the lane change to the maximum speed d ₃ (n), d ₇ (n) is the shortest forward inter-vehicle distance, d ₈ (n) is the shortest rear inter-vehicle distance, d ₉ (n) is the shortest lane change time, and d ₁₀ (n) is the longest lane change time.

In addition, the real environment observation data 122 stored in the first storage unit 120 is identified based on data such as acceleration and velocity observed from each mobile object in the real environment, and each mobile object to be observed. This is data relating to values corresponding to the moving body motion characteristic parameters d _i (i = 1,..., 10).

In addition, the moving body motion characteristic parameter reference value information 123 stored in the first storage unit 120 is a movement in which the reference value rd _i of the moving body motion characteristic parameter d _i (i = 1,..., 10) is set as one set. This is information regarding the body motion characteristic parameter reference value set ↑ rd.

  The moving body motion characteristic parameter reference value set ↑ rd may be determined based on each value included in the real environment observation data 122.

For example, the average value, median value, or mode value of the maximum acceleration d ₁ included in the actual environment observation data 122 is adopted as the reference value rd ₁ of the maximum acceleration of the moving object included in the moving body motion characteristic parameter reference value set ↑ rd. Can be done.

  Further, the moving body motion characteristic parameter reference value set ↑ rd may be determined by the user based on the respective values included in the real environment observation data 122.

(Parameter selection allowable area determination process)
Next, parameter selection allowable region determination processing executed by the first arithmetic processing unit will be described.

  First, an outline of the parameter selection allowable region determination process will be described.

  In this process, a plurality of mobile body motion characteristic sample parameter sets ↑ ds, which are a plurality of samples of the mobile body motion characteristic parameter set, and a plurality of surrounding object motion characteristic sample parameter sets, which are a plurality of samples of the surrounding object motion characteristic parameter set. ↑ e are used respectively.

  The mobile body motion characteristic sample parameter set ↑ ds is a parameter set representing the motion characteristics of one mobile body a (hereinafter referred to as “own mobile body a”). On the other hand, the surrounding object motion characteristic sample parameter set ↑ e is the other moving body b (b = 1,..., N ≠ b ≠ a) which is an object around the moving body a (hereinafter referred to as another moving body b). .) Is a parameter set representing the motion characteristics.

  Then, for each set (↑ ds, ↑ e) of the moving body motion characteristic sample parameter set ↑ ds and the surrounding object motion characteristic sample parameter set ↑ e, from a predetermined start time t = 0 to an end time t = T. Simulated motions (↑ ds, ↑ e) of the own mobile body a and the other mobile body b are generated.

  In other words, with respect to one moving body motion characteristic sample parameter set ↑ ds, simulated motions (↑ ds, ↑) of the own mobile body a and the other mobile body b from a predetermined start time t = 0 to an end time t = T. e) are generated for the number of surrounding object motion characteristic sample parameter sets ↑ e.

  Each moving body motion characteristic sample parameter set ↑ ds is evaluated based on the generated simulated motions (↑ ds, ↑ e). Then, based on the evaluation of each mobile body motion characteristic sample parameter set ↑ ds, a selection allowable region of the mobile body motion characteristic parameter set ↑ ds of the entire set of the mobile body motion characteristic parameter set ↑ ds is determined.

  Hereinafter, the parameter selection allowable region determination process will be described in more detail with reference to FIG.

  The first moving body simulated motion generation unit 111 sets the moving body motion model of the own moving body a based on the initial setting information 121 (FIG. 3 / STEP002).

  Here, the initial setting information 121 includes information related to the moving body motion model of the mobile body a and information related to the surrounding object motion characteristic sample parameter set ↑ e.

  In the present embodiment, as described above, the first moving body simulated movement generation unit 111 uses the optimal speed model and the lane change model specified by the initial setting information 121 as the moving body movement model of the own moving body a. Set as. Instead, for example, an accelerator driver model and a brake driver model described in Patent Document 1 may be set as the moving body motion model of the own moving body a.

  The first moving body simulated exercise generation unit 111 reads the moving body motion characteristic parameter reference value set ↑ rd from the moving body motion characteristic parameter reference value information 123 from the first storage unit 120 (FIG. 3 / STEP004).

  The first moving body simulated motion generation unit 111 performs a predetermined calculation on each value of the moving body motion characteristic parameter reference value set ↑ rd, for example, as shown in FIG. 4A in a two-dimensional manner. A plurality of moving body motion characteristic sample parameter sets ↑ ds, which are samples of the characteristic parameter set ↑ d, are generated (FIG. 3 / STEP006).

  The first moving body simulated exercise generation unit 111 reads real environment observation data (FIG. 3 / STEP008).

  The first moving body simulated motion generation unit 111 determines the outer edge OE of the parameter selection allowable region of the moving body motion characteristic parameter based on the real environment observation data, for example, as shown in two dimensions in FIG. 4B (FIG. 3 / (STEP010).

  More specifically, the first moving body simulated motion generation unit 111 has a predetermined ratio (for example, 90/5) of the total number of real environment observation data in the real environment observation data corresponding to the mobile body motion characteristic parameter set ↑ d. A region on the entire set of the moving body motion characteristic parameter set ↑ d that includes the above number of actual environment observation data is determined. The first moving body simulated motion generation unit 111 determines the outermost contour of the region on the entire set of the moving body motion characteristic parameter set ↑ d as the outer edge OE of the selection allowable region.

  Next, the first moving body simulated motion generation unit 111 sets the surrounding object motion characteristic sample parameter set ↑ es, which is a sample of the surrounding object motion characteristic parameter set ↑ e, based on the initial setting information 121 (FIG. 3 / (STEP012).

  More specifically, the 1st moving body simulated exercise | movement production | generation part 111 sets the number N-1 of the other moving bodies b which are surrounding objects as surrounding object motion characteristic sample parameter set ↑ es.

  Moreover, the 1st moving body simulation exercise | movement production | generation part 111 selects the movement characteristic of each other moving body b based on the initial setting information 121 as surrounding object movement characteristic sample parameter set ↑ es. More specifically, the first moving body simulated motion generation unit 111 selects one of the motion characteristics of equal acceleration and the motion characteristics of vibrating speed as the motion characteristics of each other moving body b. When the first moving body simulated motion generation unit 111 selects motion characteristics of equal acceleration, the first moving body simulation unit 111 sets the maximum speed and acceleration around only the target other moving body b from the plurality of sets of the maximum speed and acceleration. If you set the object motion characteristics sample parameter set ↑ es and select the motion characteristics where the speed vibrates, select only one of the speed vibration amplitude and maximum acceleration from the multiple sets of speed vibration amplitude and maximum acceleration. It is set as the surrounding object motion characteristic sample parameter set ↑ es of the target other moving body b.

  At the same time, the first moving body simulated motion generation unit 111 sets, as the motion characteristics of each of the other moving bodies b, the initial value (for example, the initial position x (b , 0) and initial speed v (b, 0)).

  In the present embodiment, the first moving body simulated motion generation unit 111 uses a moving body motion model that performs equal acceleration motion or a moving body motion model that vibrates at a predetermined amplitude when generating the simulated motion of the other mobile body b. However, in addition to or instead of this, the same movement model as the moving body movement model used by the own moving body a may be used.

  The first moving body simulated motion generation unit 111 substitutes the moving body motion characteristic parameter reference value set ↑ rd into ↑ d (a) of the following equation (11) to obtain the surrounding object motion characteristic sample parameter set ↑ es as an equation: By substituting for ↑ d (n) (n ≠ a) in (11), the own mobile body a and the other mobile body b (b = 1, 2,...) From the start time t = 0 to the end time t = T. , N. where b ≠ a.) (FIG. 3 / STEP014).

  In this specification, the simulated motion generated using the moving body motion characteristic parameter set ↑ d and the surrounding object motion characteristic parameter set ↑ e is expressed as SM (↑ d, ↑ e).

  Here, each X (t) (t = 0,... T) is a matrix represented by the above-described equation (2). X (0) is the initial position of each moving object n. The element x (n, t) (n = 1,... N) of each X (t) (t = 1,... T) is obtained from the above-described equation (1).

  In other words, the first moving body simulated exercise generating unit 111 converts the moving body model f into each moving body n (n = 1,..., N at each time from the start time t = 0 to the end time t = T. As a simulated motion SM (↑ rd, ↑ es), the time-series position ↑ x (n of each moving body n from the start time t = 0 to the end time t = T is applied. , T) (n = 1,..., N, including a).

  Subsequently, the first moving body simulated motion generation unit 111 sets the moving body motion characteristic sample parameter set ↑ ds generated in FIG. 3 / STEP006 as the value of the moving body motion characteristic parameter set ↑ d (a) of the own mobile body a. Is set (FIG. 3 / STEP016).

  Subsequently, the first moving body simulated motion generation unit 111 uses the above formula (10) and formula (11) for the set moving body motion characteristic sample parameter set ↑ ds, from the start time t = 0 to the end time. Simulated motion SM (↑ ds, ↑ es) of the moving body a and other moving body b up to t = T is generated (FIG. 3 / STEP018).

  The evaluation determination unit 112 determines whether or not the mobile body motion characteristic sample parameter set ↑ ds satisfies a predetermined condition based on the simulated motion SM (↑ ds, ↑ es) generated in FIG. 3 / STEP020).

  More specifically, the evaluation determination unit 112 performs the simulated motion SM (↑ rd, ↑ es) for the moving body motion characteristic parameter reference value set ↑ rd generated in FIG. 3 / STEP014, and the simulated motion in FIG. 3 / STEP018. The similarity RV with SM (↑ ds, ↑ es) is calculated. The similarity RV is, for example, the reciprocal of the distance in the parameter space between each simulated motion SM (↑ rd, ↑ es) and SM (↑ ds, ↑ es).

  Then, the evaluation determination unit 112 determines whether or not the similarity RV is equal to or higher than a predetermined similarity.

  Moreover, the evaluation determination part 112 determines whether the own mobile body a is contacting the other mobile body b based on the simulation exercise | movement SM (↑ ds, ↑ es) produced | generated by FIG.

  When the similarity RV is equal to or higher than the predetermined similarity and the mobile object a is not in contact with the other mobile object b, the evaluation determination unit 112 determines the label L (↑ ds, ↑ es) is set to 1 (valid). On the other hand, when the similarity RV is less than the predetermined similarity, or when the own mobile body a comes into contact with the other mobile body b, the evaluation determination unit 112 determines the label L (↑ ds, Set ↑ es) to 0 (invalid).

  Instead of or in addition to these conditions, the evaluation determination unit 112 determines that the acceleration or inter-vehicle distance distribution of the moving object a in the simulated motion SM (↑ ds, ↑ es) is a predetermined distribution, simulated motion The minimum value of TTC (Time To Collation) between the moving body “a” and the other moving body “b” in SM (↑ ds, ↑ es) is not less than a predetermined value, and the density of each moving object does not form a traffic jam. When the condition consisting of at least one of the following is satisfied, the label of the moving body motion characteristic sample parameter set ↑ ds is set to 1 (valid), and when the condition is not satisfied, the label L (↑ ds , ↑ es) may be set to 0 (invalid).

  The evaluation determination unit 112 determines whether or not the unexecuted mobile body motion characteristic sample parameter set ↑ ds exists for the surrounding object motion characteristic sample parameter set ↑ es set in FIG. 3 / STEP012 ( FIG. 3 / STEP022).

  If the determination result is affirmative (FIG. 3 / STEP022... YES), the first moving body simulated motion generation unit 111 is not evaluated as the moving body motion characteristic parameter set ↑ d (a) of the own moving body a. The moving body motion characteristic sample parameter set ↑ ds is set (FIG. 3 / STEP016), and the processing from FIG. 3 / STEP018 is executed.

  By repeating these processes, as shown in FIG. 4C, each moving body motion inside the outer edge OE of the selection allowable region with respect to the surrounding object motion characteristic sample parameter set ↑ es set in FIG. 3 / STEP012 Labeling for each of the characteristic sample parameter sets ↑ ds is completed. In FIG. 4C, the moving body motion characteristic sample parameter set ↑ ds where the label L (↑ ds, ↑ es) is 1 (valid) is represented by ◯, and the moving body where the label L (↑ ds, ↑ es) is 0 (invalid). The motion characteristic sample parameter set ↑ ds is indicated by x.

  When the determination result is negative (FIG. 3 / STEP022... NO), the first moving body simulated motion generation unit 111 determines whether or not an unused surrounding object motion characteristic sample parameter set ↑ es exists ( FIG. 3 / STEP024).

  When the determination result is affirmative (FIG. 3 / STEP024... YES), the first moving body simulated motion generation unit 111 uses the unused ambient object motion characteristic sample parameter set ↑ as the ambient object motion characteristic parameter set ↑ e. es is set (FIG. 3 / STEP012), and the processing from FIG. 3 / STEP014 is executed.

  In this way, by performing the processing of FIG. 3 / STEP014 and subsequent steps, each moving body motion characteristic sample parameter set ↑ ds is labeled for each surrounding object motion characteristic sample parameter set ↑ es. In other words, each moving body motion characteristic sample parameter set ↑ ds is labeled as many as the number of surrounding object motion characteristic sample parameter sets ↑ es.

  When the determination result is negative (FIG. 3 / STEP024... NO), the selection allowable region determination unit 113 determines the label L for each of the moving body motion characteristic sample parameter set ↑ ds and its surrounding object motion characteristic sample parameter set ↑ es. Based on (↑ ds, ↑ es), as shown in FIG. 4E, a simulated motion satisfying the predetermined condition of FIG. A first region R1 including a moving body motion characteristic parameter set ↑ d and a second region R2 including a moving body motion characteristic parameter set ↑ d that cannot generate a simulated motion satisfying a predetermined condition with a predetermined probability or more ( FIG. 3 / STEP026).

  More specifically, the selection permissible area determination unit 113 performs simulated motion SM (↑ for each of the moving body motion characteristic sample parameter sets ↑ ds corresponding to the different surrounding object motion characteristic sample parameter sets ↑ es. Among the total number of ds, ↑ es), the simulated exercise SM (↑ ds, ↑ es) satisfying a predetermined condition (label L (↑ ds, ↑ es) is 1 (valid) and the simulated exercise SM (↑ ds, ↑ es) ) Is calculated.

  Then, the selection allowable region determination unit 113 has a ratio of the number of simulated exercises SM (↑ ds, ↑ es) whose label L (↑ ds, ↑ es) is 1 (valid) equal to or greater than a predetermined rate (for example, 90%). A moving body in which the first teacher signal TS1 is given to a certain moving body motion characteristic sample parameter set ↑ ds1, and the ratio of the number of simulated motion SM (↑ ds, ↑ es) whose label is 1 (valid) is less than a predetermined ratio The second teacher signal TS2 is given to the motion characteristic sample parameter set ↑ ds2.

  As a result, for example, the first teacher signal TS1 or the second teacher signal TS2 is given to each of the moving body motion characteristic sample parameter sets ↑ ds. In FIG. 4D, the mobile body motion characteristic sample parameter set ↑ ds1 to which the first teacher signal TS1 is attached is indicated by a white square, and the mobile body motion characteristic sample parameter set ↑ ds2 to which the second teacher signal TS2 is attached is indicated by a black square. Show.

  Then, the selection permissible region determination unit 113 performs a support vector machine (Support Vector Machine, based on the mobile body motion characteristic sample parameter set ↑ ds1, ↑ ds2 to which the first teacher signal TS1 or the second teacher signal TS2 is given. By using a supervised classification method such as SVM), the entire set of the moving body motion characteristic parameter set ↑ d is simulated motion (Fig. 4E) that satisfies the predetermined condition of Fig. 3 / STEP 020 with a predetermined probability or more as shown in Fig. 4E. 3 / STEP 020, a first region R1 including a value of a moving body motion characteristic parameter set ↑ d that can generate a simulated motion in which a label is valid with a predetermined probability when evaluated in the same manner as STEP 020, and a predetermined probability or more In FIG. 3 / STEP020, the parameters of the moving body motion characteristics that cannot generate the simulated motion that satisfies the predetermined conditions The data is classified into the second region R2 including the value of the data set ↑ d.

  Note that the first region R1 and the second region R2 may be classified using an unsupervised classification method such as k-means instead of a supervised classification method such as a support vector machine, a principal component analysis, and the like. The first region R1 and the second region R2 may be classified using this classification method.

  The selection allowable region determination unit 113 determines the first region R1 as the selection allowable region of the moving body motion characteristic parameter d for generating the simulated motion of the mobile body, and information on the selection allowable region (for example, the boundary of the selection allowable region) Is stored in the storage unit 120 as the selection allowable region information 124 (FIG. 3 / STEP028).

  Note that the selection allowable region determination unit 113 further determines other conditions, and a moving body motion for generating a simulated motion of the moving body in a region where the predetermined region determined by the other conditions and the first region R1 overlap. It may be determined as a selection allowable region of the characteristic parameter set ↑ d.

(Movement simulation process for moving objects)
Next, referring to FIG. 5, the second moving body simulated motion generation unit 211 calculates the moving body motion model and the moving body motion characteristic parameter set ↑ d included in the selection allowable region based on the selection allowable region information 124. A moving body simulated motion generation process for generating simulated motions of a plurality (N) of moving bodies n (n = 1,..., N) in the virtual space will be described.

  This process can be used, for example, for generating a virtual traffic environment for verifying the verification of the automatic driving algorithm of the vehicle.

The second moving body simulated exercise generating unit 211 performs initial setting (FIG. 5 / STEP 102). More specifically, the second moving body simulated exercise generation unit 211 receives, from the second storage unit 220 as an initial setting, the number N of moving bodies, the simulated exercise generation time T, the initial position of each moving body n, and each moving body n. Initial setting information 221 regarding the initial velocity and the moving body motion model f _n of each moving body n is read.

Second moving member simulated motion generator 211, based on the information related to the mobile motion model f _n of the each mobile n included in the initial setting, for each of the plurality of mobile n, sets the mobile motion model f _n ( FIG. 5 / STEP 104). In the present embodiment, the above-described optimum speed model and lane change model are set as the same moving body model for all the moving bodies n.

The second moving body simulated motion generation unit 211 reads the selection allowable region information 124 corresponding to each of the moving body motion models f _n set to the moving body n from the first storage unit 120 of the parameter classification device 100, and the second It memorize | stores in the memory | storage part 220 (FIG. 5 / STEP106). In the present embodiment, since the same optimal speed model and lane change model are set for all the mobile bodies n, the second mobile body simulated motion generation unit 211 corresponds to the optimal speed model and the lane change model. The selection allowable area information 124 is read and stored in the second storage unit 220.

The second mobile body simulated motion generation unit 211 is included in the selection permissible region for each of the mobile bodies n based on the selection permissible region information 124 corresponding to each of the mobile body motion models f _n set for the mobile body n. The moving body motion characteristic parameter set ↑ d (n) is selected (FIG. 5 / STEP 108).

The second moving body simulated motion generation unit 211 performs initial setting, a moving body motion characteristic parameter set ↑ d (n) generated for each moving body n, and the moving body motion set for each of the moving bodies n. A simulated motion SM (D) of each moving body n from the start time t = 0 to the end time t = T is generated using the model f _n and the above equations (10) and (11) (FIG. 5). / STEP110).

(Operational effects of the first embodiment)
Next, with reference to FIG.6 and FIG.7, the effect of this embodiment is demonstrated.

  6A to 6F show that the number of vehicles is 20, the simulated motion generation time is 40 seconds, the initial positions are equally spaced, the initial speed is 60 km / h, and the optimal speed model is used for all moving objects. It is a graph which shows the position or speed of a moving body when the simulated movement SM (D) of each moving body n is produced | generated by the said moving body simulated exercise | movement production | generation process.

  6A to 6C are graphs in which the horizontal axis represents time and the vertical axis represents the position of each moving body.

  6D to 6F are graphs in which the horizontal axis represents time and the vertical axis represents the speed of each moving body.

  6A and 6D are graphs in the case where the mobile body motion characteristic parameter set described in Non-Patent Document 1 is used as the mobile body motion characteristic parameter set of each mobile body.

  From FIG. 6A and FIG. 6D, when the mobile body motion characteristic parameter set described in Patent Document 1 is used as the mobile body motion characteristic parameter set of each mobile body, each mobile body n (n = 1,..., 20) is one. It can be seen that simulated motions are generated that move at equal intervals at various speeds.

  6B and 6E are graphs in the case of using the mobile body motion characteristic parameter set generated based on the selection allowable region information 124 as the mobile body motion characteristic parameter set of each mobile body n.

  From FIG. 6B and FIG. 6E, when the mobile body motion characteristic parameter set generated based on the selection allowable region information 124 is used as the mobile body motion characteristic parameter set of each mobile body n, each mobile body n is not necessarily uniform. It can be seen that a simulated motion is generated in which the moving body n does not move at a uniform speed and the interval of the moving body n is not equal.

  6C and 6F are mobile body motion characteristic parameter sets generated based on the selection allowable region information 124 as the mobile body motion characteristic parameter set of each mobile body n, and the mobile body motion characteristics of FIGS. 6B and 6E. It is a graph at the time of using the moving body motion characteristic parameter set different from a parameter set.

  From FIG. 6C and FIG. 6F, when the mobile body motion characteristic parameter set is used as the mobile body motion characteristic parameter set of each mobile body n, the distance between the mobile bodies n is approximately equal, but is not necessarily uniform. It can be seen that a simulated motion is generated that does not move.

  7A to 7D show that the number of vehicles is 20, the simulated motion generation time is 40 seconds, the initial positions are equally spaced, the initial speed is 60 km / h, and the lane change model is set for all moving bodies n. When the simulated motion of each moving body n is generated by the moving body simulated motion generation process, the distance between the front and rear vehicles before and after the lane change is taken as the horizontal axis, and the frequency of occurrence of each inter-vehicle distance It is a graph which shows as a vertical axis | shaft.

  In each graph, the white bar graph indicates the frequency of appearance of each inter-vehicle distance when the simulated motion of each moving body n is generated by the moving body simulated motion generation process, and the black bar graph is obtained from the actual environment measurement data The frequency of occurrence of each inter-vehicle distance is shown.

  As can be seen from FIGS. 7A to 7D, the tendency of the frequency of appearance of each inter-vehicle distance when the simulated motion of each moving body n is generated by the moving body simulated motion generation process is obtained from the real environment measurement data by the black bar graph. This almost coincides with the frequency of appearance of each inter-vehicle distance.

  As described above, according to the selection permissible area determination process of the present embodiment, the mobile body motion characteristic parameter set ↑ d is generated using the selection permissible area information 124 generated by the process, and the mobile body motion characteristic parameter set is generated. By generating simulated motion SM (↑ d) of each moving body n using ↑ d, the behavior of the moving body in the actual traffic environment is reproduced while the behavior of the moving body in the actual traffic environment is reproduced. A simulated motion of a moving object can be generated easily and efficiently.

(Other embodiments)
The mobile body motion characteristic parameter reference value set ↑ rd of the first embodiment is obtained from the actual environment measurement data, and the mobile body motion characteristic parameter reference value set on which a driver who performs a calm operation (operation with small acceleration / deceleration) is boarded. ↑ Moving body motion characteristics parameter reference value set on which a driver performing rough driving (driving with large acceleration / deceleration) is obtained from rd1 and actual environment measurement data, and moving bodies according to driving characteristics of various drivers A motion characteristic parameter reference value set may be employed.

  Here, taking the above moving body motion characteristic parameter reference value set ↑ rd1, ↑ rd2 as an example, satisfaction and non-satisfaction of predetermined conditions in FIG. 3 / STEP020 when using a plurality of mobile body motion characteristic parameter reference value sets It supplements about the judgment of.

  The evaluation determination unit 112 includes a simulated motion SM (↑ ds, ↑ es) for the mobile body motion characteristic sample parameter set ↑ ds, a simulated motion SM (↑ rd1, ↑ es) for the mobile body motion characteristic parameter reference value set ↑ rd1, and And the simulated motion SM (↑ ds, ↑ es) for the moving body motion characteristic sample parameter set ↑ ds and the simulated motion SM (↑ rd2, ↑ es) for the mobile body motion characteristic parameter reference value set ↑ rd2 The similarity RV2 is calculated.

  When any one of the similarity RV1 and the similarity RV2 is equal to or higher than the predetermined similarity, the evaluation determination unit 112 sets the moving body motion characteristic sample parameter set ↑ ds to FIG. It is determined that the predetermined condition in is satisfied. The evaluation determination unit 112 determines that the mobile body motion characteristic sample parameter set ↑ ds does not satisfy the predetermined condition in FIG. 3 / STEP 020 when either the similarity RV1 or the similarity RV2 is not equal to or higher than the predetermined similarity.

  In addition, as the mobile body motion characteristic parameter reference value set ↑ rd of the first embodiment, the mobile body motion characteristic parameter reference value set in Saitama, the mobile body motion characteristic parameter reference value set in Tokyo, etc. acquired from the actual environment measurement data, etc. A mobile body motion characteristic parameter reference value set corresponding to the geographical mobile body motion characteristics may be employed.

  Further, as the mobile body motion characteristic parameter reference value set ↑ rd of the first embodiment, the morning mobile body motion characteristic parameter reference value set, the evening mobile body motion characteristic parameter reference value set, etc. acquired from the actual environment measurement data, etc. A mobile body motion characteristic parameter reference value set corresponding to the temporal motion characteristics of the mobile body may be employed.

  In the first embodiment, the evaluation determination unit 112 determines whether each of the moving body motion characteristic sample parameter sets ↑ ds satisfies a predetermined condition as a set of classification target parameters. It may be determined whether the surrounding object motion characteristic sample parameter set ↑ es satisfies a predetermined condition.

  In the first embodiment, the mobile body motion characteristic parameter set is defined as ↑ d (n) and is a time-invariant parameter. However, the mobile body motion characteristic parameter set is defined as ↑ d (n, t), It may be a time variable parameter.

(Second Embodiment)
Next, a parameter classification device 100 ′ according to the second embodiment of the present invention will be described with reference to FIGS.

  As shown in FIG. 8, in the parameter classification device 100 ′ of the present embodiment, the first arithmetic processing unit 110 ′ includes an attention classification target parameter determination unit 114, and the first storage unit 120 ′ includes attention classification target parameter information 125. Is different from the first embodiment in that it is stored. Portions common to the first embodiment are denoted by the same reference numerals and description thereof is omitted.

(Attention classification target parameter determination processing)
With reference to FIG. 9, the attention classification object parameter determination process which the 1st moving body simulation exercise | movement production | generation part 111, the evaluation determination part 112, and the attention classification object parameter determination part 114 performs is demonstrated.

  The processes of FIG. 9 / STEP002 to STEP024 performed by the first moving body simulated exercise generating unit 111 and the evaluation determining unit 112 are the same as those of FIG. 3 / STEP002 to STEP024 in the selection allowable region process, and thus the description thereof is omitted.

The attention classification target parameter determination unit 114 defines the domain for each mobile body motion characteristic parameter d _i based on the label for each mobile body motion characteristic sample parameter set ↑ ds given from the evaluation determination unit 112 in FIG. Statistical information is calculated (FIG. 9 / STEP 226).

For example, the attention classification target parameter determination unit 114 includes the effective label L (↑) assigned to each moving body motion characteristic sample parameter set ↑ ds in the domain of the maximum acceleration d ₁ among the moving body motion characteristic parameters d _i. The distribution of the ratio RL (↑ ds) of ds, ↑ es) is calculated.

Further, for example, interest classified parameter determination unit 114, maximum acceleration valid label per unit acceleration in the d ₁ of domain L (↑ ds, ↑ es) the variation of the average value of the ratio RL (↑ ds) of calculate.

The attention classification target parameter determination unit 114 has a relatively large influence on the satisfaction determination of a predetermined condition among the mobile body motion characteristic parameters d _i based on the statistical information about the respective mobile body motion characteristic parameters d _i ( The attention classification target parameter, which is a parameter that is presumed to be highly sensitive to the satisfaction of the predetermined condition, is determined (FIG. 9 / STEP 228).

For example, the ratio RL (↑ ds) of the effective labels L (↑ ds, ↑ es) given to each moving body motion characteristic sample parameter set ↑ ds with respect to the maximum acceleration d ₁ out of the mobile body motion characteristic parameters d _i When the distribution degree of the distribution is the largest, the attention classification target parameter determination unit 114 determines the maximum acceleration as the attention classification target parameter.

For example, among the mobile motion characteristic parameters d _i, for a plurality of mobile motion characteristic parameters d _i, the ratio of each mobile motion characteristics sample parameter set ↑ ds effective granted to label L (↑ ds, ↑ es) When the distribution degree of the distribution of RL (↑ ds) is greater than or equal to a predetermined value, the attention classification target parameter determination unit 114 may determine the plurality of moving body motion characteristic parameters d _i as a target classification target parameter set.

Further, for example, each mobile motion characteristics sample parameter set ↑ ds effective granted to label L (↑ ds, ↑ es) for the percentage of RL (↑ ds), the change per unit in the domain of maximum acceleration d ₁ When the amount is equal to or larger than the predetermined change amount, the attention classification target parameter determination unit 114 may determine the maximum acceleration as the attention classification target parameter.

  The attention classification target parameter determination unit 114 stores information on the determined attention classification target parameter (or attention classification target parameter set) in the first storage unit 120 'as attention classification target parameter information 125 (FIG. 9 / STEP 230).

(Operational effect of the second embodiment)
According to the second embodiment, an attention classification target parameter that is presumed to have a relatively large influence on satisfaction with a predetermined condition (high sensitivity to satisfaction with a predetermined condition) is determined as an attention classification target parameter. .

  For example, in STEP 006 of the selection allowable region determination process (FIG. 3), the first moving body simulated motion generation unit 111 reduces the granularity of the sample value generation of the target classification target parameter, thereby increasing the sensitivity to the satisfaction determination of the predetermined condition. From the viewpoint of appropriateness, the granularity of the moving body motion characteristic parameter is appropriately adjusted.

  As a result, the probability that the first region classified in STEP 026 of the selection allowable region determination process (FIG. 3) includes a moving body motion characteristic parameter capable of generating a simulated motion that meets the evaluation criterion is further increased.

  As a result, the simulated motion generated from the moving body motion characteristic parameter value selected from the selection permissible region determined in STEP 028 of the selection permissible region determination process (FIG. 3) becomes more likely to meet the evaluation criteria.

  In addition, for example, the second moving body simulated motion generation unit 211 generates the moving body motion characteristic parameter value more finely for the target classification target parameter in STEP 106 of the moving body simulated motion generation process (FIG. 5). Various traffic environments can be reproduced.

(Third embodiment)
Next, the third embodiment of the present invention will be described using the parameter classification device 100 and the actual moving body 300 (corresponding to the “target moving body” of the present invention) shown in FIG. 10 as examples.

  Portions common to the first embodiment and the second embodiment are denoted by the same reference numerals and description thereof is omitted.

  In the present embodiment, the actual moving body 300 is a moving body of the same type (for example, a four-wheeled vehicle) as the type used in the first embodiment.

  The actual moving body 300 includes a third arithmetic processing unit 310, a third storage unit 320, an external environment recognition device 331 that recognizes environment information around the actual moving body 300, and a drive mechanism that drives the actual moving body 300. 332, a braking mechanism 333 that brakes the actual moving body 300, and a steering mechanism 334 that steers the actual moving body 300.

  The third arithmetic processing unit 310 includes a moving body motion determination unit 311, an external environment recognition unit 312, and an operation control unit 313.

  The third arithmetic processing unit 310 is configured by an arithmetic processing unit of one or a plurality of computers mounted on the actual mobile unit 300 reading and executing a mobile unit operation control program stored in a storage unit of the computer. . The device constituting the third arithmetic processing unit 310 corresponds to the “moving body motion generation device” of the present invention.

  The external environment recognition unit 312 is an external device such as an imaging device such as a camera connected to the computer, an object detection device such as a laser for detecting an object existing outside, an inter-vehicle communication device or a road-vehicle communication device. By analyzing information about the external environment obtained through the environment recognition device 331, the surrounding object motion characteristic parameter set is parameterized and recognized.

  Further, the operation control unit 313 outputs an operation instruction signal to each mechanism (each device) such as the drive mechanism 332, the braking mechanism 333, and the steering mechanism 334 connected to the computer, so that each mechanism (each Device).

  The mobile body motion determination unit 311, the external environment recognition unit 312, and the motion control unit 313 are configured to execute a mobile body motion control process described later.

  The third storage unit 320 includes storage devices such as a main storage device, a secondary storage device, and an auxiliary storage device. Instead, the third storage unit 320 may be configured by a database server provided by a cloud service.

  The third storage unit 320 is configured to store the mobile body motion model information 321 stored in advance and the selection allowable region information 124 acquired from the parameter classification device 100.

  Note that the moving body motion characteristic parameter included in the selection allowable region information 124 is a parameter used for the moving body motion model included in the moving body motion model information 321.

  The computers mounted on the parameter classification device 100 and the actual moving body 300 may be configured by physically separate hardware (computers), respectively, or a part (for example, a storage device) may be shared and the rest (for example, arithmetic processing) Apparatus) may be configured by separate hardware, or all may be configured by the same hardware.

  When the first storage unit 120 of the parameter classification device 100 and the third storage unit 320 of the actual mobile body 300 are configured by separate hardware, the actual mobile body 300 is connected to the parameter classification device 100 via a network such as the Internet. The selection allowable area information 124 is acquired from the first storage unit 120 and stored in the third storage unit 320.

(Moving object operation control processing)
Next, with reference to FIG. 11, the moving body operation | movement control process performed by the real moving body 300 is demonstrated.

  The mobile body motion determination unit 311 sets the mobile body motion model of the actual mobile body 300 based on the mobile body motion model information 321 stored in the third storage unit 320 (FIG. 11 / STEP 302).

  More specifically, the moving body motion determination unit 311 uses an optimal speed model and a lane that are the same moving body motion model as the moving body motion model used in the parameter classification device 100 as the moving body motion model of the actual moving body 300. Set the change model.

  Based on the selection allowable region information 124 stored in the third storage unit 320, the mobile body motion determination unit 311 selects any mobile body motion characteristic parameter included in the selection allowable region from the entire set of mobile body motion characteristic parameters. The set ↑ d is selected as the moving body motion characteristic parameter set ↑ d of the actual moving body 300 (FIG. 11 / STEP 304).

  The external environment recognizing unit 312 receives an object m (m = 1,..., M, M is an actual movement) such as a moving body existing around the real moving body 300 as information about the external environment via the external environment recognizing device 331. The position ↑ x (m) of the number of objects existing around the body 300 is sequentially recognized (FIG. 12 / STEP 306). For example, the external environment recognition unit 312 recognizes the position ↑ x (m) of the object m existing around the actual moving body 300 by analyzing the captured image acquired from the imaging device serving as the external environment recognition device 331. To do.

  The mobile body motion determination unit 311 includes the mobile body motion model set in FIG. 11 / STEP 302, the mobile body motion characteristic parameter set ↑ d set in FIG. 11 / STEP 304, and the actual motion recognized in FIG. 11 / STEP 306. Based on the position ↑ x (m) of the object m existing around the body 300, using the equations (1) to (9), the current state of the actual moving body 300 as the movement of the actual moving body 300 Next, the position after unit time Δt is sequentially determined (FIG. 11 / STEP308).

  The motion control unit 313 is an accelerator for moving to the position so that the actual position of the actual moving body 300 follows the position after the unit time Δt from the current position of the actual moving body 300 determined in FIG. Control information such as the opening degree, the braking operation amount, and the steering angle is calculated, and the operation of each mechanism is controlled based on the calculated control information (FIG. 11 / STEP 310).

  After the processing of FIG. 11 / STEP 310, the third arithmetic processing unit 310 repeatedly executes the processing of FIG. 11 / STEP 306 and subsequent steps.

(Operational effect of the third embodiment)
According to the real moving body 300 of the third embodiment, based on the selection permissible area information generated by the parameter classification device 100, any motion characteristic parameter included in the selection permissible area in the entire set of mobile body motion characteristic parameters. The set ↑ d is selected as the moving body motion characteristic parameter set of the actual moving body 300. Here, the selection allowable region is a region including a moving body motion characteristic parameter set that can generate a simulated motion that conforms to a predetermined evaluation criterion, or a partial region thereof. Therefore, by determining the motion of the actual moving body 300 using the moving body motion characteristic parameter set included in the selection allowable region, the probability that the determined motion of the actual moving body 300 matches a predetermined evaluation criterion is relatively high. Become expensive.

  That is, according to the real moving body 300 of the third embodiment, it is possible to easily and efficiently generate the movement of the real moving body 300 having a relatively high probability of meeting a predetermined evaluation criterion.

  DESCRIPTION OF SYMBOLS 100 ... Parameter classification | category apparatus, 111 ... 1st moving body simulated exercise | movement production | generation part, 112 ... Evaluation determination part, 113 ... Selection permissible area | region determination part, R1 ... 1st area | region, R2 ... 2nd area | region.

Claims (12)

A first motion that generates a simulated motion of the moving body using a moving body motion model that is described by a parameter set that includes a plurality of parameters including a predetermined type of classification target parameter and that represents the motion of the moving body. A moving body simulated motion generation unit;
The first moving body simulated motion generation unit is caused to generate a simulated motion of the moving body using each of a plurality of sample parameter sets that are a plurality of samples of the parameter set, and the plurality of sample parameter sets Based on the simulated motion generated corresponding to each of the plurality of sample parameter sets, the classification target parameter sample value that is the value of the classification target parameter included in each of the plurality of sample parameter sets is a simulated motion that conforms to a predetermined evaluation criterion. An evaluation determination unit that determines whether or not the value can be generated;
In accordance with the determination result of the evaluation determination unit, the entire set of classification target parameters, a first region including the classification target parameters capable of generating a simulated motion that conforms to the evaluation criterion, and a simulation that conforms to the evaluation criterion Classifying the movement into the second region including the classification target parameter that cannot be generated;
A parameter classification device comprising: a selection permissible region determining unit that determines the first region or a partial region thereof as a selection permissible region of the classification target parameter for generating a simulated motion of the moving body. The parameter classification device according to claim 1, wherein
The parameter set includes a moving body motion characteristic parameter indicating a motion characteristic of the moving body and a surrounding object motion characteristic parameter indicating a motion characteristic of an object existing around the moving body. The parameter classification device according to claim 2, wherein
The parameter set includes the classification target parameter and a classification non-target parameter that is a non-classification parameter,
The evaluation determination unit
The plurality of sample parameter sets are a plurality of samples in which the classification target parameter sample values are the same and the classification non-target parameter sample values that are values of the classification non-target parameters included in the sample parameter set are different from each other. Including parameter sets,
Of the total number of the simulated exercises generated corresponding to each of the plurality of sample parameter sets including the same classification target parameter sample values, the ratio of the number of simulated exercises satisfying a predetermined condition is equal to or greater than a predetermined ratio. The parameter classification apparatus characterized by determining that the same classification target parameter sample value is a value capable of generating a simulated motion that conforms to a predetermined evaluation criterion. The parameter classification device according to claim 3, wherein
The parameter classification apparatus, wherein the predetermined condition is a condition that the moving body does not contact another object In the parameter classification device according to any one of claims 2 to 4,
The parameter classification apparatus, wherein the classification target parameter is the moving body motion characteristic parameter. In the parameter classification device according to any one of claims 1 to 5,
The evaluation determination unit includes a classification target parameter criterion which is a value of a classification target parameter serving as a reference and a simulated movement of the mobile unit generated by using the classification target parameter sample value in the first moving target simulated movement generation unit. If the similarity with the simulated motion of the mobile body generated by the first mobile body simulated motion generation unit using a value is equal to or higher than a predetermined similarity, the classification target parameter sample value is a predetermined value A parameter classification device that determines that a simulated motion that meets the evaluation criteria is a value that can be generated. The parameter classification device according to claim 6, wherein
The classification target parameter reference value is set based on observation data obtained by observing movement of a moving object in a real environment. In the parameter classification device according to claim 6 or 7,
The parameter classification apparatus, wherein the classification target parameter reference value is a plurality of different classification target parameter reference values. In the parameter classification device according to any one of claims 6 to 8,
The selection allowable region determining unit
Based on the observation data obtained by observing the movement of the moving object in the real environment, the existence area of the classification target parameter in the entire set of classification target parameters,
An apparatus for classifying parameters, wherein an overlapping area between the classification target parameter existing area and the first area is determined as the selection allowable area. In the parameter classification device according to any one of claims 1 to 9,
The classification that has a relatively large influence on the conformity determination of the simulated motion in which the change in the value of the classification target parameter is generated based on the definition area for each value of the classification target parameter in the selection allowable area A parameter classification device comprising: a target classification target parameter determination unit that determines a target parameter as a target classification target parameter.   A simulated motion of the moving body is generated using the classification target parameter included in the selection allowable region determined by the parameter classification device according to any one of claims 1 to 10 and the moving body motion model. A moving body simulated exercise generating apparatus comprising: a second moving body simulated exercise generating unit. An external environment recognition unit for recognizing information on the external environment as a value of the surrounding object motion characteristic parameter;
A mobile body motion characteristic parameter included in the selection allowable region determined by the parameter classification device according to claim 5, the mobile body motion model, and information on the external environment recognized by the external environment recognition unit. A mobile body motion determination unit that determines the motion of the target mobile body;
And a motion control unit for controlling the operation of the device of the target mobile body so as to realize the motion of the target mobile body determined by the mobile body motion determination unit. apparatus.