DE102022102118A1 - Method and apparatus for generating trajectories on a given route for multiple vehicles with given performance specifications - Google Patents

Abstract

The invention provides a method for generating trajectories on a given route for a plurality of vehicles with given performance specifications, characterized in that the performance specifications of the vehicles (M1, M2, M3) used are selected; loading road information related to the route; driving scenario specifications are defined based on the driving requirements; using the driving scenario specifications, the route is divided into fully specified segments and solved independently by forward simulation of the driving requirements, thereby obtaining boundary conditions for any remaining under-specified segments of the route; for each of the insufficiently specified segments, an optimal control problem, based on a physical model of the vehicle (M1, M2, M3) and constrained by the driving requirements, of minimizing the performance criterion is formulated and solved by means of non-linear programming; and the resulting trajectories (1, 2, 3) for each vehicle (M1, M2, M3) are merged at safe pass-by positions along the route and an execution order of the vehicles (M1, M2, M3) is determined by flow shop scheduling. The invention further provides a corresponding data processing device, a corresponding computer program and a corresponding medium.

Description

The present invention relates to a method for generating trajectories on a given route for multiple vehicles with given performance specifications. The invention further relates to a corresponding data processing device, a corresponding computer program and a corresponding medium.

State of the art

The goal of automating endurance testing is to enable safe and repeatable durability testing of vehicles, increase track utilization, reduce labor costs, and eliminate human factors such as driver fatigue.

The prior art includes commercial systems for the automated repetition of recorded trajectories on a test track. However, with these known approaches, only the automated repetition of a previously traveled trajectory is supported. The definition of artificial maneuvers is typically limited to simple building blocks of constant velocity or constant acceleration segments. The definition process requires a large amount of user input, has limited ability to capture the general behavior of an experienced test driver, and relies on the operator to ensure the safety and feasibility of the test scenario. Therefore, the trajectory must first be driven by a human driver, which limits the flexibility of the solution when creating new test scenarios. In addition, with these known approaches, collisions between the vehicles in the plan are only prevented reactively during the test execution, which can invalidate the previous test progress. This makes the manual process of test plan definition not scalable for large fleets of automated test vehicles, since the operator is responsible for preventing unwanted interference between the test vehicles and their individual trajectories.

US20190235516A1 uses a spline-curve-based path optimization scheme, followed by spline-curve-based velocity trajectory optimization.

U.S. 2019/079523AA , U.S. 2019/079528AA , U.S. 2019/086925AA , U.S. 2019/086930AA and U.S. 2019/317509AA refer to control systems for autonomous vehicles that include means for generating trajectories for a given road or track segment, calculating and using the vehicle's optimal control, speed or trajectory.

Other known approaches to automated endurance testing do not allow automated rescheduling and rescheduling in the event of adding or removing a test vehicle.

Disclosure of Invention

The invention is based on the problem - given a desired path, vehicle performance specifications (such as tire grip or engine power) and test scenario requirements (such as constant acceleration, constant speed or speed-limited segments) - of planning a complete speed trajectory that satisfies the given constraints and minimizes the performance criteria in terms of execution time, required acceleration, and so on.

To this end the invention provides a method for generating trajectories on a given route for a plurality of vehicles with given performance specifications, a corresponding data processing device, a corresponding computer program and a corresponding medium according to the independent claims.

The proposed optimization-based framework offers automated generation of speed trajectories based on driving requirements and performance specifications with full control over driving style via parameterization and differs in its flexibility from the existing automated route driving concepts.

Advantageously, the decomposition of the problem into sub-problems of planning and scheduling enables parallelization and scalability for multiple vehicles and joint use of routes.

A preferred embodiment is applicable for speed planning on curved roads taking into account tire grip, required braking distance, available braking power and any trade-offs between performance and comfort.

Further embodiments of the invention are specified in the dependent claims, which allow, among other things, an intuitive coordination between aggressive and cautious driving style and a systematic treatment of driving dynamics and driving restrictions, as evidenced by an additional check of the driving requirements Feasibility before optimization, a physically based vehicle motion model included in trajectory optimization, or a systematic consideration of tire traction limits as optimization constraints.

character list

• The only 1 shows schematically an application of the invention.

implementation of the invention

• Zunächst erzeugt eine globale Planungsvorrichtung (GPL) eine Trajektorie über ein gegebenes Streckensegment, indem sie ein Problem der optimalen Steuerung für jedes Streckensegment mit den aus den Testspezifikationen bestimmten Geschwindigkeitsrandbedingungen löst, was in einer Geschwindigkeitstrajektorie für jedes Fahrzeug und jedes Streckensegment resultiert. Die GPL stellt sicher, dass jedes Fahrzeug die geforderte Geschwindigkeitstrajektorie ausführt, wenn dies möglich ist, und erkennt ansonsten die Nicht-Machbarkeit der Anforderungen.

According to the invention, to enable scalability to multiple vehicles, the trajectory planning problem is decomposed into independent sub-problems using test scenario specifications. The proposed trajectory planning framework consists of two main modules:

• First, a global planner (GPL) generates a trajectory over a given route segment by solving an optimal control problem for each route segment with the speed constraints determined from the test specifications, resulting in a speed trajectory for each vehicle and each route segment. The GPL ensures that each vehicle executes the required speed trajectory when it can, and otherwise recognizes the non-feasibility of the requirements.

Second, a Scheduling Device (SCH) combines individual vehicle trajectories for route sharing using execution scheduling to avoid collisions and minimize total execution time. Here, vehicle scheduling is solved by the exclusive assignment of route segments to individual vehicles and formulated as a flow shop scheduling problem, which results in a schedule of execution of vehicle trajectories (orders) assigned to route segments (machines), including the order of the vehicles and the overtaking sequence and minimizing the overall duration of the test plan execution ("cycle time").

As can be seen from the above, the proposed architecture is highly scalable due to the decomposition of planning and scheduling, its suitability for parallelization (independent optimization for each route segment and vehicle) and the possible server or cloud-based deployment.

• Als erste Eingabe wird die Strecke in Form von Wegedaten beschrieben, die auf einer Karte basieren oder separat in Abhängigkeit von geometrischen Beschränkungen der Straßenbegrenzungen wie ein minimaler Krümmungsweg geplant werden können. Der gewünschte Weg kann als Wegepunkte oder als eine Kombination aus Ausgangsposition, Orientierung und Krümmung als Funktion der zurückgelegten Strecke gegeben werden.

To generate the trajectories, a method according to the invention relies on several inputs:

• As a first input, the route is described in the form of path data, which can be based on a map or planned separately depending on geometric restrictions of the road boundaries such as a minimum curvature path. The desired path can be given as waypoints or as a combination of starting position, orientation and curvature as a function of distance traveled.

Optionally, road gradient and angle of departure, tire-to-road adhesion and overtaking opportunities along the route can also be specified.

If only waypoints are specified by their position, the curvature is reconstructed using optimization based on a kinematic vehicle model, minimizing the weighted sum of the squares of the position error (fit), the required curvature, and the curvature derivative.

As a second input, driving requirements are specified. For example, constant speed over a segment of the route, ABS braking or full throttle acceleration over a given portion of the route, speed ranges that must be maintained, and similar test maneuvers can be specified.

As a third input, vehicle performance is specified in terms of input power (e.g., engine power, efficiency, and vehicle mass), losses (e.g., rolling resistance or drag), tire peak lateral or longitudinal force, and other model-based physical constraints.

As a fourth input, a specific performance criteria weight can be specified. Such tuning parameters make it possible to obtain behaviors ranging from a race-style trajectory with minimal time to a "cautious" trajectory with minimal energy consumption, and include any combination of the above such as penalized acceleration trajectories and their time derivatives.

• Erstens werden basierend auf dem verwendeten Fahrzeug Leistungsspezifikationen ausgewählt und nach Erfordernis modifiziert.
• Zweitens werden Straßeninformationen wie Krümmung, Haftung oder zulässiger Geschwindigkeitsbereich geladen und modifiziert, falls erforderlich.

From the software perspective, the proposed procedure includes the following steps:

• First, performance specifications are selected based on the vehicle used and modified as necessary.
• Secondly, road information such as curvature, grip or permissible speed scope loaded and modified if necessary.

Third, fixed driving scenario specifications can be defined in terms of constant speed or constant acceleration segments. The feasibility of each specification is checked by comparison to a maximum speed profile obtained using current road conditions and a backwards simulation, requesting further input in the case of non-feasibility.

Fourth, using the driving scenario specifications, only the fully specified path segments are selected and resolved independently through forward simulation of the driving requirements. The results provide the boundary conditions for the remaining, insufficiently specified route segments.

Fifth, for each remaining under-specified segment, a velocity trajectory is determined by optimization that minimizes the weighted performance criterion of total execution time and power consumption.

Sixth, the resulting trajectories for each vehicle are merged between possible overtaking places, and flow shop scheduling is performed to determine the execution order of the vehicles. The trajectories of the vehicles waiting at the safe overtaking locations are modified accordingly, for example by increasing the standing time.

The orderly execution of these steps results in the complete overall plan or schedule for all vehicles over a prescribed number of track laps.

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• US20190235516A1 [0004]
• US2019079523AA [0005]
• US2019079528AA [0005]
• US2019086925AA [0005]
• US2019086930AA[0005]
• US2019317509AA [0005]

Claims (10)

Method for generating trajectories (1, 2, 3) on a given route for a number of vehicles (M1, M2, M3) with given performance specifications while observing given driving requirements and a given performance criterion, characterized in that - the performance specifications of the vehicles used (M1 , M2, M3) are selected, - road information related to the route is loaded, - based on the driving requirements, driving scenario specifications are defined for each of the vehicles used (M1, M2, M3), - using the driving scenario specifications, the route is divided into fully specified segments for each of the vehicles used (M1, M2, M3) and solved independently by forward simulation of the driving requirements, thereby obtaining boundary conditions for any remaining under-specified segments of the route, - for each of the under-specified segments, an optimal control problem, based on a physical model of the vehicle (M1, M2, M3) and constrained by the driving requirements, minimizing the performance criterion and solved by non-linear programming and - the resulting trajectories (1, 2, 3) for each vehicle (M1, M2, M3 ) are merged at safe drive-by positions along the route and an execution order of the vehicles (M1, M2, M3) is determined by flow shop scheduling. procedure after claim 1 , characterized in that - the road information relates to position, curvature, adhesion and permissible speed range. procedure after claim 1 or 2 , characterized in that - the driving scenario specifications relate to segments of the route that are to be driven at constant speed or acceleration. procedure after claim 3 , characterized in that - after defining the driving scenario specifications, the feasibility of each specification is checked by comparing the specification with a maximum speed profile under current road conditions on the relevant segments, in case of non-feasibility further input is requested. procedure after claim 4 , characterized in that - the maximum speed profile is obtained by reverse simulation based on the performance specifications of each vehicle and the current road conditions. Procedure according to one of Claims 1 until 5 , characterized in that - the performance criterion is a weighted sum of the total execution time and the energy required to follow the trajectories (1, 2, 3). Procedure according to one of Claims 1 until 6 , characterized in that , - the trajectories (1, 2, 3) of the vehicles (M1, M2, M3) at the pass-by positions are modified according to the execution order, such as prolonged standing still. Data processing device that has means for performing the method according to one of Claims 1 until 7 has. Computer program adapted to carry out the method according to any one of Claims 1 until 7 . Computer-readable medium with the program stored on it claim 9 .

Images