CN117565887A - Service recommendation method, vehicle-mounted terminal and vehicle - Google Patents
Service recommendation method, vehicle-mounted terminal and vehicle Download PDFInfo
- Publication number
- CN117565887A CN117565887A CN202311577044.6A CN202311577044A CN117565887A CN 117565887 A CN117565887 A CN 117565887A CN 202311577044 A CN202311577044 A CN 202311577044A CN 117565887 A CN117565887 A CN 117565887A
- Authority
- CN
- China
- Prior art keywords
- service recommendation
- data
- model
- driving state
- state data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 32
- 238000012549 training Methods 0.000 claims description 26
- 238000012545 processing Methods 0.000 claims description 14
- 238000004140 cleaning Methods 0.000 claims description 11
- 238000012795 verification Methods 0.000 claims description 10
- 238000011156 evaluation Methods 0.000 claims description 8
- 238000005070 sampling Methods 0.000 claims description 7
- 230000002159 abnormal effect Effects 0.000 claims description 6
- 230000001133 acceleration Effects 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000012360 testing method Methods 0.000 claims description 5
- 238000000638 solvent extraction Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 5
- 230000006870 function Effects 0.000 description 6
- 230000006399 behavior Effects 0.000 description 3
- 206010063385 Intellectualisation Diseases 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000013450 outlier detection Methods 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000000546 chi-square test Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000000556 factor analysis Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/08—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to drivers or passengers
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
- B60W50/08—Interaction between the driver and the control system
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/0895—Weakly supervised learning, e.g. semi-supervised or self-supervised learning
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W40/00—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models
- B60W40/08—Estimation or calculation of non-directly measurable driving parameters for road vehicle drive control systems not related to the control of a particular sub unit, e.g. by using mathematical models related to drivers or passengers
- B60W2040/089—Driver voice
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W50/00—Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
- B60W2050/0001—Details of the control system
- B60W2050/0043—Signal treatments, identification of variables or parameters, parameter estimation or state estimation
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60W—CONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
- B60W2540/00—Input parameters relating to occupants
- B60W2540/21—Voice
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Automation & Control Theory (AREA)
- Evolutionary Computation (AREA)
- Mechanical Engineering (AREA)
- Data Mining & Analysis (AREA)
- Transportation (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Health & Medical Sciences (AREA)
- Human Computer Interaction (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a service recommendation method, a vehicle-mounted terminal and a vehicle, wherein driving state data is input into a pre-trained service recommendation model to obtain a service recommendation strategy, service recommendation is performed based on the service recommendation strategy, feedback information input by a user based on the service recommendation is obtained to update the service recommendation model based on the feedback information, so that the current driving state is understood and judged through the pre-trained service recommendation model, corresponding intelligent service recommendation is made, and the service recommendation model is updated based on the information fed back by the user to improve personalized service recommendation effects.
Description
Technical Field
The present invention relates to the field of vehicle technologies, and in particular, to a service recommendation method, a vehicle-mounted terminal, and a vehicle.
Background
At present, in the vehicle-mounted terminal, multiple types of services such as navigation, entertainment and life can be provided for users generally, however, the traditional vehicle-mounted terminal often needs to manually design and write fixed rules to realize service recommendation, but the personalized service recommendation effect of the method is poor.
Disclosure of Invention
Aiming at the defects in the prior art, the embodiment of the invention provides a service recommendation method, a vehicle-mounted terminal and a vehicle.
In order to solve the above technical problems, an embodiment of the present invention provides a service recommendation method, including:
acquiring driving state data; wherein the driving state data includes at least one of a driver voice instruction, vehicle state information, and driving environment information;
inputting the driving state data into a pre-trained service recommendation model to obtain a service recommendation strategy;
performing service recommendation based on the service recommendation policy;
acquiring feedback information input by a user based on the service recommendation;
and updating the service recommendation model based on the feedback information.
Preferably, the training step of the service recommendation model includes:
collecting driving state data, wherein the driving state data comprises a driver voice instruction, vehicle state information and driving environment information;
data cleaning is carried out on the driving state data;
carrying out data coding on the driving state data after data cleaning;
inputting the driving state data after data coding into a service recommendation model for training;
evaluating the service recommendation model according to the evaluation index;
and when the evaluation result does not meet the self-learning performance requirement, adjusting or retraining the service recommendation model.
Preferably, the training of the driving state data after the data encoding is input into a service recommendation model specifically includes:
constructing a service recommendation model;
inputting the driving state data after data encoding into a service recommendation model to obtain an output result of the service recommendation model;
importing an output result of the service recommendation model into a self-supervision learning module; wherein the self-supervision learning module is composed of a mirror image of the service recommendation model;
and importing the user feedback information into a self-supervision learning module so that the self-supervision learning module performs self-supervision and learning growth based on the user feedback information and checks the output result of the service recommendation model.
Preferably, after the user feedback information is imported into the self-supervised learning module, so that the self-supervised learning module performs self-supervision and learning growth based on the user feedback information, and verifies an output result of the service recommendation model, the method further includes:
when the output result of the service recommendation model passes the verification, outputting the output result of the service recommendation model;
and when the output result of the service recommendation model fails to pass the verification, updating the data label, adding the learning result parameter, and returning to the step of inputting the driving state data after the data coding into the service recommendation model to obtain the output result of the service recommendation model.
Preferably, the constructing a service recommendation model specifically includes:
and constructing a service recommendation model by adopting a GPT-4 model, a T5 model, a RoBERTa model, a BERT model or an XLnet model.
Preferably, the step of data cleaning the driving state data specifically includes:
performing data preprocessing on the driving state data;
carrying out data sampling on the preprocessed driving state data;
carrying out data deduplication on the driving state data after data sampling;
carrying out standardized processing on driving state data after data deduplication;
carrying out abnormal value processing on the driving state data after the standardized processing;
carrying out data reconstruction and feature selection on the driving state data processed by the abnormal value;
dividing a data set of driving state data after data reconstruction and feature selection;
and storing the divided driving state data.
As a preferred solution, the dividing the data set by the driving state data after the data reconstruction and the feature selection specifically includes:
the driving state data after data reconstruction and feature selection are divided into a training set, a verification set and a test set.
Preferably, the vehicle state information includes at least one of a vehicle speed, a steering angle, a braking state and an acceleration, and the service recommendation policy includes a driving recommendation suggestion based on the vehicle state information;
the driving environment information includes at least one of a road condition, a traffic sign, a traffic light state, and a front vehicle running condition, and the service recommendation policy includes a driving assistance recommendation suggestion based on the driving environment information.
In order to solve the same technical problem, an embodiment of the present invention further provides a vehicle-mounted terminal, including:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the service recommendation method.
In order to solve the same technical problems, the embodiment of the invention also provides a vehicle, which comprises the vehicle-mounted terminal.
Compared with the prior art, the invention has the beneficial effects that: the embodiment of the invention provides a service recommendation method, a vehicle-mounted terminal and a vehicle, wherein driving state data is input into a pre-trained service recommendation model to obtain a service recommendation strategy, service recommendation is performed based on the service recommendation strategy, feedback information input by a user based on the service recommendation is obtained to update the service recommendation model based on the feedback information, so that the current driving state is understood and judged through the pre-trained service recommendation model, corresponding intelligent service recommendation is made, and the service recommendation model is updated based on the information fed back by the user to improve personalized service recommendation effect.
Drawings
FIG. 1 is a flow chart of a service recommendation method according to an embodiment of the present invention;
fig. 2 is a training schematic diagram of a service recommendation model in an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Fig. 1 is a flowchart illustrating a service recommendation method according to an embodiment of the invention.
The service recommendation method provided by the embodiment of the invention comprises the following steps:
step S101, driving state data is obtained; wherein the driving state data includes at least one of a driver voice instruction, vehicle state information, and driving environment information;
step S102, inputting the driving state data into a pre-trained service recommendation model to obtain a service recommendation strategy;
step S103, service recommendation is carried out based on the service recommendation strategy;
step S104, obtaining feedback information input by a user based on the service recommendation;
step S105, updating the service recommendation model based on the feedback information.
In the embodiment of the invention, the driving state data is input into the pre-trained service recommendation model to obtain the service recommendation strategy, the service recommendation is performed based on the service recommendation strategy, then the feedback information input by the user based on the service recommendation is obtained to update the service recommendation model based on the feedback information, so that the current driving state is understood and judged through the pre-trained service recommendation model, the corresponding intelligent service recommendation is made, and the service recommendation model is updated based on the information fed back by the user to improve the personalized service recommendation effect.
In an embodiment of the present invention, the vehicle state information includes at least one of a vehicle speed, a steering angle, a braking state, and an acceleration, and the service recommendation policy includes a driving recommendation suggestion based on the vehicle state information; the driving environment information includes at least one of a road condition, a traffic sign, a traffic light state, and a front vehicle running condition, and the service recommendation policy includes a driving assistance recommendation suggestion based on the driving environment information.
Illustratively, a voice command regarding the driver: the voice command of the driver is used for the system to judge the intention of the user and execute corresponding operation, and the voice command of the driver is converted into a text or command which can be understood by a machine through a voice recognition technology, so that the interaction with the vehicle-mounted artificial intelligent assistant is realized, for example, the driver can tell the assistant that the user needs to navigate to a certain destination through the voice command, adjust the temperature of an air conditioner, play music and the like. Regarding vehicle state information: the vehicle state information includes vehicle speed, steering angle, braking state, acceleration, etc., which can be used to monitor the running state of the vehicle, determine driving behavior and driving safety, and the assistant can provide feedback and advice of the driver's behavior, such as reminding the driver of overspeed, overheat of the vehicle, etc., by analyzing the vehicle state information. Regarding driving environment information: the driving environment information comprises road conditions, traffic signs, traffic light states, driving conditions of vehicles in front and the like, and the information can be used for providing driving auxiliary functions, such as traffic jam reminding, front collision early warning and the like, and through monitoring the driving environment in real time, a driver can be helped to make more intelligent decisions, and driving safety is improved.
In the embodiment of the invention, the bottom layer of the service recommendation model is the generalization of the general large model, so that a user can input data into the model at will during use. By acquiring the voice command of the driver, the vehicle state, the driving environment and other information, the model can understand and judge the current driving state, so that corresponding decisions can be made or driving assistance functions can be provided. Illustratively: 1. voice command of driver: the driver can inform the requirements of the vehicle-mounted artificial intelligence assistant through voice instructions, such as navigation destination, temperature adjustment, music switching and the like, and the model can conduct voice recognition on the voice instructions of the driver and convert the voice instructions into instructions which can be understood by the machine. 2. Vehicle state information: the vehicle state information includes vehicle speed, steering angle, braking state, acceleration, etc., and the model can use the information to determine the current driving state, for example, determine whether the vehicle speed is too high, determine whether sudden braking exists, etc. 3. Driving environment information: the driving environment information includes road conditions, traffic signs, traffic light states, driving conditions of the front vehicle, etc., and the model can evaluate the safety and fluency of the current driving environment by using the information so as to provide corresponding driving assistance functions, such as traffic jam reminding, front collision early warning, etc.
In practice, the output of the model depends on what information the user has specifically entered into the model, and different answers will be generated according to the different inputs, but in the in-vehicle scenario, the common outputs of the model are: 1. providing voice driving advice: optimal route planning, proper vehicle speed advice, energy-efficient driving skills, etc. 2. The driving assistance function, the model can provide corresponding driving assistance functions such as automatic cruise control, lane keeping assistance, traffic jam detour and the like according to the requirements of users and the current driving situation. 3. Passenger chatting and life assistant service, for example, the user can send out voice as follows: the model carries out corresponding reply suggestions on 'help me take out', 'help me order restaurant', 'chat with me', and the like.
In the embodiment of the invention, the training step of the service recommendation model comprises the following steps:
step S11, collecting driving state data, wherein the driving state data comprises a driver voice instruction, vehicle state information and driving environment information;
step S12, data cleaning is carried out on the driving state data;
step S13, data encoding is carried out on the driving state data after data cleaning;
s14, inputting the driving state data after data encoding into a service recommendation model for training;
step S15, evaluating the service recommendation model according to the evaluation index;
and S16, when the evaluation result does not meet the self-learning performance requirement, adjusting or retraining the service recommendation model.
In an implementation, the driving state data includes a driver voice command, vehicle state information including at least one of a vehicle speed, a steering angle, a braking state, and an acceleration, and driving environment information including at least one of a road condition, a traffic sign, a traffic light state, and a forward vehicle running condition. The source of driving state data can be obtained by means of questionnaires, recording driver behaviors, monitoring vehicle states in real time, importing existing data of an automobile host factory and the like. In order to guarantee the quality of the data, the collected data should have a certain representativeness and diversity.
In addition, the data cleaning is to remove noise data and error data so as to improve the quality and accuracy of model training. The data coding is to digitally code data such as voice and text in order to convert the original data into a data format which can be processed by a computer, and the process is also called feature extraction, and different models need different feature extraction methods, for example, for text data, the data can be coded in a word vector, character vector and other modes. After model training is completed, the model needs to be evaluated, and various evaluation indexes such as accuracy, recall, F1 value, AUC and the like can be adopted for evaluating the performance of the model, and if the self-learning performance of the model is not strong enough, the model needs to be finely tuned or retrained so as to ensure the accuracy of the model. Finally, the trained model can be deployed in an actual application scene, namely a vehicle-mounted environment, and the model can be optionally embedded into vehicle-mounted equipment directly or deployed in a mode of providing an API interface through cloud service and the like.
In some embodiments, the step S12 "data-cleaning the driving status data" specifically includes:
performing data preprocessing on the driving state data;
carrying out data sampling on the preprocessed driving state data;
carrying out data deduplication on the driving state data after data sampling;
carrying out standardized processing on driving state data after data deduplication;
carrying out abnormal value processing on the driving state data after the standardized processing;
carrying out data reconstruction and feature selection on the driving state data processed by the abnormal value;
dividing a data set of driving state data after data reconstruction and feature selection;
and storing the divided driving state data.
Exemplary, the data set partitioning of the driving state data after data reconstruction and feature selection specifically includes: the driving state data after data reconstruction and feature selection are divided into a training set, a verification set and a test set.
In an embodiment of the present invention, regarding data preprocessing: the raw data is imported into an operational format, such as CSV, JSON, etc., in which some preprocessing steps, such as data format conversion, missing value padding, outlier handling, etc., are required. Regarding data sampling: and on the premise of ensuring that the distribution and the quantity of the data meet the actual conditions, a small part of data is randomly selected for cleaning so as to save time and calculation resources. Regarding data deduplication: for repeated data, a deduplication operation is required to avoid repeated computation and training, and this step may use existing deduplication algorithms, such as hash deduplication algorithms, rank deduplication algorithms, cluster deduplication algorithms, and the like. Regarding data normalization: for numerical data, a normalization operation is required to ensure that the comparison between different features is reasonable. Regarding data outlier processing: outliers can cause greater interference with model training, and outlier detection and processing is required, and this step can be processed using statistical methods, clustering methods, outlier detection algorithms, and the like. Regarding data reconstruction feature selection: for unstructured data, data reconstruction and feature selection are needed to extract important features and remove irrelevant features, and in specific applications, feature selection can be performed by using methods such as principal component analysis, factor analysis, chi-square test and the like. Regarding data set partitioning: the cleaned data set is divided into a training set, a verification set and a test set for model training, evaluation and testing. Regarding data storage: the cleaned data needs to be stored for subsequent model training and application, for example, a database, cloud storage and the like can be used for storage. According to the embodiment of the invention, through data cleaning, noise data and error data are removed, so that the quality and accuracy of model training are improved.
In some embodiments, the step S14 "training the driving state data encoded with the data in the service recommendation model" specifically includes:
step S141, constructing a service recommendation model;
step S141, driving state data after data coding is input into a service recommendation model to obtain an output result of the service recommendation model;
step S141, importing the output result of the service recommendation model into a self-supervision learning module; wherein the self-supervision learning module is composed of a mirror image of the service recommendation model;
step S141, importing the user feedback information into a self-supervision learning module, so that the self-supervision learning module performs self-supervision and learning growth based on the user feedback information, and checking the output result of the service recommendation model.
In some embodiments, the step S141 of "building a service recommendation model" specifically includes:
and constructing a service recommendation model by adopting a GPT-4 model, a T5 model, a RoBERTa model, a BERT model or an XLnet model.
It should be noted that, while the traditional machine learning modes, such as decision trees, support vector machines, neural networks, etc., the embodiment of the present invention selects the most recent version of the pre-trained large model, such as GPT-4/T5/RoBERTa/BERT/XLNet, etc., as the system core module to perform tasks and human intent understanding.
In some embodiments, after the step S141 of introducing the user feedback information into the self-supervised learning module to enable the self-supervised learning module to perform self-supervision and learning growth based on the user feedback information and verify the output result of the service recommendation model, the method further includes:
when the output result of the service recommendation model passes the verification, outputting the output result of the service recommendation model;
and when the output result of the service recommendation model fails to pass the verification, updating the data label, adding the learning result parameter, and returning to the step of inputting the driving state data after the data coding into the service recommendation model to obtain the output result of the service recommendation model.
Referring to fig. 2, in the implementation, a stored data set is trained through a pre-trained large model of the latest version such as GPT-4/T5/RoBERTa/BERT/XLNet and the like, and training results are imported into a self-supervision learning module, the module is composed of a pre-trained large model mirror image, self-supervision and learning growth is performed through collecting user feedback, and output results of the large model are checked, the contents checked by the self-supervision learning module are directly output, data labels are updated through the non-passing results, learning result parameters are added, and data are imported into the data set again, so that learning and growth of the large model are facilitated. The embodiment of the invention promotes the self-growth of the model by adding the method for forming the self-supervision learning module by using the pre-training large model mirror image, and can further solve the problem that artificial intelligence is realized by manually designing and writing rules in the prior art.
Correspondingly, the embodiment of the invention also provides a vehicle-mounted terminal, which comprises:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the service recommendation method.
It should be appreciated that the processor may be a central processing unit (Central Processing Unit, CPU), but may also be other general purpose processors, digital signal processors (Digital Dignal Processing, DSP), application specific integrated circuits (Application Specific Sntegrated Circuit, ASIC), field programmable gate arrays (Field Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or the like. A general purpose processor may be a microprocessor or any conventional processor or the like. It is noted that the processor may be a processor supporting an advanced reduced instruction set machine (advanced RISC machines, ARM) architecture.
Correspondingly, the embodiment of the invention also provides a vehicle, which comprises the vehicle-mounted terminal. In particular, in order to recognize the voice signal and play the voice, a voice recognition device and an audio play device may be provided, and the voice recognition device and the audio play device may be integrated in the vehicle-mounted terminal or may be provided separately from the vehicle-mounted terminal, for example. In addition, in order to facilitate the user to input feedback information, an input device, such as a touch screen, may be provided, and the touch screen may be integrated in the vehicle-mounted terminal, or may be provided separately from the vehicle-mounted terminal. In the implementation, the vehicle-mounted terminal and the mobile phone and other devices can be in communication connection, so that more service functions can be realized, for example, a song of the mobile phone can be accessed to the vehicle-mounted terminal for playing and the like.
Compared with the prior art, the embodiment of the invention has at least one of the following advantages:
1. individualizing: in the prior art, it is often necessary to manually design and write rules to achieve personalized dialog. However, this method requires human intervention and cannot cover all cases, so that the individualization effect is limited. By adopting the new model training method, the model can automatically learn the language habits and preferences of different users from a large amount of data, thereby generating more personalized dialogue.
2. Intelligent: in the prior art, manual writing and maintenance are required for the rules, and the rules can only cover a few cases, so that real intellectualization is difficult to realize. By adopting a humanized model training method, the model can learn the complexity and potential relation of human language from a large amount of data, thereby realizing true intellectualization.
3. And (3) naturalizing: in the prior art, due to the complexity of human language, all situations cannot be considered in advance, so that the constraint and control are often required to be performed by using preset rules. But this approach tends to result in language models that behave hard and unnatural. By adopting the model training method, the model can better understand the complexity of human language, so that more natural and smooth dialogue is generated.
The foregoing is merely a preferred embodiment of the present invention, and it should be noted that modifications and substitutions can be made by those skilled in the art without departing from the technical principles of the present invention, and these modifications and substitutions should also be considered as being within the scope of the present invention.
Claims (10)
1. A service recommendation method, comprising:
acquiring driving state data; wherein the driving state data includes at least one of a driver voice instruction, vehicle state information, and driving environment information;
inputting the driving state data into a pre-trained service recommendation model to obtain a service recommendation strategy;
performing service recommendation based on the service recommendation policy;
acquiring feedback information input by a user based on the service recommendation;
and updating the service recommendation model based on the feedback information.
2. The service recommendation method of claim 1, wherein the training step of the service recommendation model comprises:
collecting driving state data, wherein the driving state data comprises a driver voice instruction, vehicle state information and driving environment information;
data cleaning is carried out on the driving state data;
carrying out data coding on the driving state data after data cleaning;
inputting the driving state data after data coding into a service recommendation model for training;
evaluating the service recommendation model according to the evaluation index;
and when the evaluation result does not meet the self-learning performance requirement, adjusting or retraining the service recommendation model.
3. The service recommendation method as claimed in claim 2, wherein the data-encoded driving state data is input into a service recommendation model for training, specifically comprising:
constructing a service recommendation model;
inputting the driving state data after data encoding into a service recommendation model to obtain an output result of the service recommendation model;
importing an output result of the service recommendation model into a self-supervision learning module; wherein the self-supervision learning module is composed of a mirror image of the service recommendation model;
and importing the user feedback information into a self-supervision learning module so that the self-supervision learning module performs self-supervision and learning growth based on the user feedback information and checks the output result of the service recommendation model.
4. The service recommendation method of claim 3, further comprising, after said importing user feedback information into a self-supervised learning module to cause said self-supervised learning module to perform self-supervision and learning growth based on said user feedback information and to verify an output result of a service recommendation model:
when the output result of the service recommendation model passes the verification, outputting the output result of the service recommendation model;
and when the output result of the service recommendation model fails to pass the verification, updating the data label, adding the learning result parameter, and returning to the step of inputting the driving state data after the data coding into the service recommendation model to obtain the output result of the service recommendation model.
5. The service recommendation method as claimed in claim 3, wherein said constructing a service recommendation model specifically comprises:
and constructing a service recommendation model by adopting a GPT-4 model, a T5 model, a RoBERTa model, a BERT model or an XLnet model.
6. The service recommendation method according to claim 2, wherein the data cleansing of the driving state data specifically includes:
performing data preprocessing on the driving state data;
carrying out data sampling on the preprocessed driving state data;
carrying out data deduplication on the driving state data after data sampling;
carrying out standardized processing on driving state data after data deduplication;
carrying out abnormal value processing on the driving state data after the standardized processing;
carrying out data reconstruction and feature selection on the driving state data processed by the abnormal value;
dividing a data set of driving state data after data reconstruction and feature selection;
and storing the divided driving state data.
7. The service recommendation method according to claim 6, wherein the data set partitioning of the driving state data after data reconstruction and feature selection specifically includes:
the driving state data after data reconstruction and feature selection are divided into a training set, a verification set and a test set.
8. The service recommendation method according to any one of claims 1 to 7, wherein the vehicle state information includes at least one of a vehicle speed, a steering angle, a braking state, and an acceleration, and the service recommendation policy includes a driving recommendation suggestion based on the vehicle state information;
the driving environment information includes at least one of a road condition, a traffic sign, a traffic light state, and a front vehicle running condition, and the service recommendation policy includes a driving assistance recommendation suggestion based on the driving environment information.
9. A vehicle-mounted terminal, characterized by comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the service recommendation method of any one of claims 1-9.
10. A vehicle comprising the in-vehicle terminal according to claim 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311577044.6A CN117565887A (en) | 2023-11-23 | 2023-11-23 | Service recommendation method, vehicle-mounted terminal and vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311577044.6A CN117565887A (en) | 2023-11-23 | 2023-11-23 | Service recommendation method, vehicle-mounted terminal and vehicle |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117565887A true CN117565887A (en) | 2024-02-20 |
Family
ID=89889578
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311577044.6A Pending CN117565887A (en) | 2023-11-23 | 2023-11-23 | Service recommendation method, vehicle-mounted terminal and vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117565887A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118194998A (en) * | 2024-03-07 | 2024-06-14 | 重庆赛力斯凤凰智创科技有限公司 | Reply generation method, device and storage medium |
-
2023
- 2023-11-23 CN CN202311577044.6A patent/CN117565887A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN118194998A (en) * | 2024-03-07 | 2024-06-14 | 重庆赛力斯凤凰智创科技有限公司 | Reply generation method, device and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11034362B2 (en) | Portable personalization | |
| US12005922B2 (en) | Toward simulation of driver behavior in driving automation | |
| DE102013223036B4 (en) | Adaptation methods for language systems | |
| CN109760585A (en) | With the onboard system of passenger traffic | |
| CN103928027A (en) | Adaptation Methods And Systems For Speech Systems | |
| CN109920410B (en) | Apparatus and method for determining reliability of recommendation based on environment of vehicle | |
| CN117565887A (en) | Service recommendation method, vehicle-mounted terminal and vehicle | |
| CN110503949B (en) | Dialogue system, vehicle having dialogue system, and dialogue processing method | |
| CN111402925A (en) | Voice adjusting method and device, electronic equipment, vehicle-mounted system and readable medium | |
| US11794676B1 (en) | Computing systems and methods for generating user-specific automated vehicle actions using artificial intelligence | |
| CN110203154A (en) | Recommended method, device, electronic equipment and the computer storage medium of vehicle functions | |
| CN115509572A (en) | Method for dynamically configuring business logic, cloud platform, vehicle and storage medium | |
| Singh | Enhancing driving experience and safety through chatgpt in vehicle systems | |
| US20240112562A1 (en) | Systems and methods for increasing the safety of voice conversations between drivers and remote parties | |
| WO2024126023A1 (en) | Computing systems and methods for generating user-specific automated vehicle actions | |
| US11900916B2 (en) | Vehicle output based on local language/dialect | |
| CN115859219A (en) | Multi-modal interaction method, device, equipment and storage medium | |
| US20210326758A1 (en) | Techniques for automatically and objectively identifying intense responses and updating decisions related to input/output devices accordingly | |
| CN115628753A (en) | Navigation voice broadcasting method and device, electronic equipment and storage medium | |
| Lin | Cross-layer system design for autonomous driving | |
| CN116168704B (en) | Voice interaction guiding method, device, equipment, medium and vehicle | |
| CN117316159B (en) | Vehicle voice control method, device, equipment and storage medium | |
| CN118092715A (en) | Vehicle system layout adjustment method, electronic device and computer readable storage medium | |
| EP3889851A1 (en) | System, method and computer program for verifying learned patterns using assis-tive machine learning | |
| CN115237440A (en) | Information processing method and device, electronic equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |