Intelligent driving model training method and device, equipment and storage medium

Abstract

Embodiments of the present application relate to the technical field of vehicles, and disclose a training method, device and equipment of an intelligent driving model and a storage medium, the method comprising: in response to a driving operation of a user in a first driving scene, obtaining first driving behavior data corresponding to the driving operation; determining a target score corresponding to the first driving behavior data according to a score reward and punishment strategy, the score reward and punishment strategy being determined based on traffic rule constraints and / or a preset path time; and if the target score is greater than a first preset score, updating an intelligent driving model deployed in a vehicle according to the first driving behavior data. The technical solution of the present application can realize training of an intelligent driving model that is more in line with an individual user on the basis of meeting the safety and compliance of an intelligent driving system.

Description

Technical Field

[0001] This invention relates to the field of vehicle technology, specifically to a training method, apparatus, device, and storage medium for an intelligent driving model. Background Technology

[0002] With the increasing prevalence of autonomous driving technology, users' personalized demands for vehicle driving behavior are growing. For example, some users prefer aggressive driving, while others prefer conservative driving.

[0003] Currently, intelligent driving systems typically use a unified algorithm model, which makes it difficult to adapt to users' diverse driving habits, leading to a decrease in users' trust in intelligent driving systems. Summary of the Invention

[0004] In view of the above problems, embodiments of the present invention provide a training method, apparatus, device and storage medium for intelligent driving models, which are used to solve the technical problem of insufficient personalization of intelligent driving models in the prior art.

[0005] According to one aspect of the present invention, a method for training an intelligent driving model is provided, the method comprising:

[0006] In response to the user's driving operation in the first driving scenario, the first driving behavior data corresponding to the driving operation is obtained;

[0007] According to the score reward and punishment strategy, the target score corresponding to the first driving behavior data is determined, and the score reward and punishment strategy is determined based on traffic rule constraints and / or preset route time.

[0008] If the target score is greater than the first preset score, the intelligent driving model deployed in the vehicle is updated based on the first driving behavior data.

[0009] According to another aspect of the present invention, a training apparatus for an intelligent driving model is provided, comprising:

[0010] The acquisition module is used to acquire first driving behavior data of the driving operation in response to the user's driving operation in the first driving scenario;

[0011] The determination module is used to determine the target score corresponding to the first driving behavior data according to the score reward and punishment strategy, wherein the score reward and punishment strategy is determined based on traffic rule constraints and / or preset path time.

[0012] The training module is used to update the intelligent driving model deployed in the vehicle based on the first driving behavior data if the target score is greater than the first preset score.

[0013] According to another aspect of the present invention, an electronic device is provided, including: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other through the communication bus;

[0014] The memory is used to store at least one executable instruction that causes the processor to perform operations such as the training method for the intelligent driving model described above.

[0015] According to another aspect of the present invention, a computer-readable storage medium is provided:

[0016] The storage medium stores at least one executable instruction that causes the training device / electronic device of the intelligent driving model to perform the operation of the training method of the intelligent driving model described above.

[0017] According to another aspect of the present invention, a computer program product is provided, including a computer program that, when executed by a processor, causes a training device / electronic device for an intelligent driving model to perform the operation of the above-described method.

[0018] This invention, in response to a user's driving operation in a first driving scenario, acquires first driving behavior data corresponding to the operation. Based on a score-based reward and punishment strategy, a target score is determined for the first driving behavior data. This strategy is based on traffic rule constraints and / or a preset path time. If the target score is greater than a first preset score, the intelligent driving model deployed in the vehicle is updated based on the first driving behavior data. This solution associates user driving behavior data with a score-based reward and punishment strategy based on traffic rules and traffic efficiency in real time. When a user's operation satisfies the target score exceeding a preset threshold, their behavior data is used to update the intelligent driving model. This ensures that all data entering the model training meets basic safety and efficiency standards. Thus, during continuous personalized optimization, the model can gradually adapt to the user's driving habits while fundamentally preventing the model from learning unsafe or inefficient driving modes. Ultimately, this improves driving comfort and acceptability while ensuring the safety and compliance of the intelligent driving system.

[0019] The above description is merely an overview of the technical solutions of the embodiments of the present invention. In order to better understand the technical means of the embodiments of the present invention and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0020] The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0021] Figure 1 A schematic diagram of the training system for the intelligent driving model provided by the present invention is shown.

[0022] Figure 2 A flowchart of the first embodiment of the training method for the intelligent driving model provided by the present invention is shown;

[0023] Figure 3 A flowchart of a second embodiment of the training method for the intelligent driving model provided by the present invention is shown;

[0024] Figure 4 A flowchart of a third embodiment of the training method for the intelligent driving model provided by the present invention is shown;

[0025] Figure 5 A flowchart of a fourth embodiment of the training method for the intelligent driving model provided by the present invention is shown;

[0026] Figure 6 A schematic diagram of an embodiment of the training device for the intelligent driving model provided by the present invention is shown;

[0027] Figure 7 A schematic diagram of an embodiment of the electronic device provided by the present invention is shown. Detailed Implementation

[0028] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0029] With the popularization of autonomous driving technology, users' personalized needs for vehicle driving behavior are increasing. For example, some users prefer aggressive driving (such as quick lane changes and acceleration), while others prefer conservative driving (such as maintaining a large following distance and slow braking).

[0030] Currently, intelligent driving systems typically employ a unified intelligent driving model, which generates standardized decision-making logic through training on large-scale real-world road data. However, the following technical problems exist:

[0031] 1) Homogenization of intelligent driving models: Existing models cannot adapt to the diverse driving habits of users, resulting in low user acceptance of intelligent driving systems;

[0032] 2) Data collection cost and security issues: Relying on data accumulated from real road tests is costly and poses security risks, making it difficult to meet the needs of large-scale personalized model training.

[0033] 3) User trust and participation issues: Users feel that the system's decision-making logic does not conform to their personal habits and lacks intuitive interaction methods to encourage users to actively participate in model optimization.

[0034] 4) Complex scenario adaptation problem: The existing model cannot handle the personalized needs of dynamic traffic scenarios (such as emergency avoidance when AEB is triggered), resulting in poor system performance in specific scenarios.

[0035] Based on the above-mentioned technical problems, the technical concept of this invention is as follows: Addressing the issue of differences in user habits, a gamified driving scenario can be designed to recreate the user's frequently used driving areas, allowing the user to simulate real driving behavior within that scenario. Driving behavior data can then be extracted. To ensure the reliability of this data, a score-based reward and punishment mechanism can be established based on traffic regulations, usual route travel time, etc., to score the driving behavior data. When the score exceeds a predetermined threshold, it is used as the training set for an intelligent driving model, thus obtaining a personalized intelligent driving model.

[0036] Based on the above technical concept, Figure 1 This invention illustrates the principle of the training system for the intelligent driving model provided by the present invention, as shown in the diagram. Figure 1 As shown, the system includes: acquisition layer, processing layer, training layer, execution layer, and cloud layer.

[0037] For the data acquisition layer, software, namely a simulation game platform, is deployed to simulate real driving scenarios and collect driving behavior data corresponding to the user's driving operations.

[0038] For the processing layer, software conditions are deployed, including: a behavior quantification module, a rule scoring module, and a dual-gated filtering algorithm; these are used to score and filter driving behavior data (blocking abnormal data), and are used as input for the training layer after all conditions are met.

[0039] For the training layer, software conditions are deployed, including: a reinforcement learning trainer and a hierarchical reward and punishment module, which are used to train the intelligent driving model and implement rewards and punishments for users;

[0040] For the execution layer, software conditions are deployed, including: a vehicle intelligent driving system, used for intelligent driving based on a trained intelligent driving model;

[0041] For the cloud layer, an over-the-air (OTA) upgrade module is deployed for remotely updating the intelligent driving model.

[0042] The above provides a brief introduction to the intelligent driving model training system involved in this invention. Based on the above schematic diagram of the system principle, the technical solution of this invention will be described in detail through specific embodiments. The executing entity of this invention is an electronic device, which may be a vehicle, a controller in a vehicle, etc. It should be noted that the following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0043] Figure 2 A flowchart illustrating a first embodiment of the training method for an intelligent driving model provided by the present invention is shown, the method being executed by an electronic device. Figure 2 As shown, the method includes the following steps:

[0044] Step 21: In response to the user's driving operation in the first driving scenario, obtain the first driving behavior data corresponding to the driving operation;

[0045] In this step, the user performs driving operations in the driving scenario corresponding to the actual or simulated driving environment (such as a real road scene recreated 1:1 in a simulation game platform deployed in the vehicle), i.e., the first driving scenario. The vehicle collects and records its steering, acceleration, braking and other operation data in the first driving scenario in real time, which are recorded as the first driving behavior data.

[0046] The first driving behavior data includes driving behavior data at various timestamps (i.e., sampling points) in the driving scenario. Each driving behavior data corresponds to a driving environment, such as road topology, traffic light timing, road congestion, and other scenario information at that timestamp.

[0047] Optionally, the first driving scenario is a real driving scenario. Accordingly, the implementation of obtaining the first driving behavior data corresponding to the driving operation can be: when the automatic emergency braking (AEB) system in the vehicle is detected to be triggered, the first driving behavior data corresponding to the driving operation is obtained.

[0048] In this implementation, when the vehicle is in a real driving scenario, and the AEB of the vehicle is detected to be triggered, the vehicle can collect and record relevant driving operation data of the user before and after the trigger, including: steering, acceleration, braking and other behaviors, as the first driving behavior data.

[0049] In one possible implementation, the first driving scenario could be the user's usual commute to work, i.e., the entire commuting scenario.

[0050] Step 22: Determine the target score corresponding to the first driving behavior data based on the score reward and punishment strategy;

[0051] The point-based reward and penalty strategy is determined based on traffic rule constraints and / or preset route time.

[0052] In this step, the user's first driving behavior data is scored based on a score-based reward and punishment strategy constructed according to preset traffic rule constraints (i.e., describing driving regulations) and / or route time (i.e., describing traffic efficiency).

[0053] For example, whether it involves running red lights, changing lanes over solid lines, yielding to pedestrians, speeding, and commuting time.

[0054] Then, a comprehensive rule score is calculated to assess the safety and compliance of the user's driving behavior.

[0055] Optionally, the score-based reward and punishment strategy may include at least one of the following:

[0056] 1) If driving behavior data indicates a violation of traffic rules, specifically running a red light, the first point will be deducted;

[0057] In this implementation, if the driving behavior data indicates a violation of traffic rules, such as running a red light, then the violation is determined based on the specific behavior, such as running a red light or repeatedly running a yellow light, and the corresponding first point is deducted from the total score.

[0058] 2) If the driving behavior data indicates an illegal lane change that violates traffic rules, a second point will be deducted;

[0059] In this implementation, if driving behavior data indicates an illegal lane change that violates traffic rules, for example, if a vehicle is detected changing lanes in a solid line area, the behavior is determined to be a violation, and the corresponding second point is deducted from the total rule score.

[0060] 3) If driving behavior data indicates a violation of the traffic rule requirement of yielding to pedestrians, a third point will be deducted;

[0061] In this implementation, if driving behavior data indicates a violation of the traffic rule constraint of yielding to pedestrians, such as failing to maintain a safe distance or failing to stop and yield at a crosswalk, it is determined as a failure to yield to pedestrians, and the corresponding third point is deducted from the total rule score.

[0062] 4) If driving behavior data indicates speeding in violation of traffic rules, the fourth point will be deducted based on the speeding percentage.

[0063] In this implementation, if the driving behavior data indicates speeding as a violation of traffic rules, a tiered judgment is made based on the proportion of the vehicle's actual speed exceeding the speed limit, such as 20%, 50%, or 100%, and the corresponding fourth points are deducted accordingly.

[0064] That is, for each speeding percentage, there is a different fourth score.

[0065] 5) If the driving behavior data indicates that the actual route time is longer than the preset route time, the fifth point will be deducted based on the overtime ratio.

[0066] In this implementation, if the driving behavior data indicates that the actual path time is greater than the preset path time (e.g., the optimal path time varies depending on the driving scenario), then a tiered judgment is made based on the overtime ratio, such as 20%, 50%, 100%, etc., and the corresponding fifth score is deducted accordingly.

[0067] That is, for each timeout percentage, there is a different fifth score.

[0068] It should be understood that in the above implementation, a total rule score can be preset, and based on the first driving behavior data and the score reward and punishment strategy, deductions are made on the total rule score to finally obtain the target score.

[0069] Step 23: If the target score is greater than the first preset score, update the intelligent driving model deployed in the vehicle based on the first driving behavior data.

[0070] In this step, if the target score is greater than the first preset score, then the user's driving operation is considered to have a rule score that is higher than the set threshold (i.e., it indicates that the safety and compliance filter has been passed), and the user's first driving behavior data is considered to be valid input.

[0071] Furthermore, by using reinforcement learning and other methods, the parameters of existing intelligent driving models that are not part of the vehicle are optimized and updated, so that the updated intelligent driving models comply with traffic rules and are more in line with the user's driving habits in subsequent intelligent driving processes.

[0072] Furthermore, after step 22, the method may also include the following possible implementations:

[0073] The first option is to control the vehicle to perform a first operation if the target score is less than or equal to the first preset score. The first operation is to instruct the user to learn traffic safety rules.

[0074] In this implementation, if the target score is less than or equal to the first preset score, it indicates that the user's driving operation has many safety hazards or violations. At this time, the safety learning mechanism corresponding to the first operation can be forcibly triggered.

[0075] For example, traffic rule education content, safe driving videos, or interactive learning modules can be pushed to the vehicle's infotainment screen or the owner's corresponding terminal to guide users to learn and confirm, thereby improving their driving safety awareness and compliance.

[0076] The second method is to send the first point to the user's account if the target score is greater than the second preset score. The first point can be used to purchase a preset object.

[0077] The second preset score is greater than the first preset score.

[0078] In this implementation, if the target score is greater than the second preset score, it indicates that the user's overall driving behavior is good and meets the requirements of safety and efficiency. At this time, a certain number of points can be issued to the user's account as a reward. These first points can be used to redeem goods, services or vehicle-related benefits in the cooperative platform or vehicle service system, thereby incentivizing users to maintain and optimize their driving habits.

[0079] The intelligent driving model training method provided in this invention obtains first driving behavior data corresponding to a user's driving operation in a first driving scenario; determines a target score corresponding to the first driving behavior data according to a score reward and punishment strategy, which is determined based on traffic rule constraints and / or preset path time; if the target score is greater than a first preset score, the intelligent driving model deployed in the vehicle is updated based on the first driving behavior data. This solution associates user driving behavior data with a score reward and punishment strategy based on traffic rules and traffic efficiency in real time. When a user's operation satisfies the target score being greater than a preset threshold, their behavior data is used to update the intelligent driving model. This ensures that the data entering the model training meets basic safety and efficiency standards. Thus, during continuous personalized optimization, it can gradually adapt to the user's driving habits and fundamentally avoid the model learning unsafe or inefficient driving modes. Ultimately, while improving driving comfort and acceptability, it ensures the safety and compliance of the intelligent driving system.

[0080] Based on the above embodiments, Figure 3 A flowchart of a second embodiment of the training method for the intelligent driving model provided by the present invention is shown. Figure 3 As shown, step 23 above may include the following steps:

[0081] Step 31: Filter the first driving behavior data according to the filtering strategy;

[0082] The filtering strategy includes: at least one driving behavior parameter and a preset value range for each driving behavior parameter;

[0083] In this step, based on the preset value range corresponding to each driving behavior parameter in the filtering strategy, the value corresponding to the driving behavior parameter of the first driving behavior data is determined to be within the corresponding preset value range, which is a reasonable range for normal data.

[0084] If a driving behavior parameter exceeds the corresponding range, the driving behavior segment associated with that parameter is marked as abnormal and filtered out. Only data with parameters within the valid range are retained, thereby ensuring that the data input into the training of the intelligent driving model is reasonable and representative.

[0085] Optionally, one possible implementation of step 31 could be:

[0086] Step 1: Quantify the first driving behavior data into multiple first driving behavior parameters;

[0087] In this implementation, the first driving behavior data can be quantified into multiple first driving behavior parameters. That is, according to the preset mapping rules, the collected original driving operations (e.g., steering wheel rotation, pedal opening and closing, etc.) are transformed into driving behavior parameters with clear physical meaning and computability.

[0088] For example, steering angular velocity, distance to the vehicle in front, lateral acceleration, and rate of change of pedal speed form a structured set of parameters, which facilitates subsequent standardized analysis and filtering.

[0089] For example, Table 1 is a schematic table of mapping rules between driving behavior data and driving behavior parameters provided by the present invention, as shown in Table 1:

[0090] Table 1:

[0091]

[0092] Step 2: Based on the filtering strategy, filter the driving behavior parameters that are not located in the corresponding preset value range among the multiple first driving behavior parameters.

[0093] In this implementation, according to the filtering strategy, driving behavior parameters that are not located in the corresponding preset value range among multiple first driving behavior parameters are filtered. That is, each quantized parameter is compared with its preset valid value range. If a parameter exceeds its reasonable range (for example, the steering angle is too large or the following distance is too close), the parameter is determined to be an outlier and its associated driving behavior segment is removed from the training dataset.

[0094] Step 32: Update the intelligent driving model deployed in the vehicle based on the filtered first driving behavior data.

[0095] In this step, the first driving behavior data, after being filtered for effectiveness, is input into the reinforcement learning trainer, compared with the decision of the original intelligent driving model, and the relevant parameters in the intelligent driving model are adjusted accordingly to form a personalized driving strategy that is more in line with the user's habits, thereby obtaining an intelligent driving model that can be applied to subsequent intelligent driving.

[0096] It should be understood that this step can be implemented on the vehicle side, i.e., the vehicle performs local updates to the intelligent driving model; or the filtered first driving behavior data can be uploaded to the cloud, where the intelligent driving model is updated and trained, and then the updated intelligent driving model is updated to the vehicle via OTA technology.

[0097] The intelligent driving model training method provided in this invention filters first driving behavior data according to a filtering strategy, which includes at least one driving behavior parameter and a preset numerical range corresponding to each driving behavior parameter. The intelligent driving model deployed in the vehicle is then updated based on the filtered first driving behavior data. This solution uses a preset numerical range for specific driving behavior parameters as a filtering strategy to screen the acquired first driving behavior data. Only data meeting the preset reasonable range requirements is used for subsequent updates to the intelligent driving model. This effectively eliminates unreasonable, unsafe, or physically inconsistent behavior data caused by misoperation, extreme scenarios, or noise interference, thereby ensuring the effectiveness and safety of the input training data for the intelligent driving model. It avoids low-quality or dangerous driving modes contaminating the intelligent driving model, ultimately improving the reliability, safety, and predictability of the personalized intelligent driving model, while reducing the risk of model oscillation or performance degradation due to data noise.

[0098] Based on the above embodiments, Figure 4 A flowchart of a third embodiment of the training method for the intelligent driving model provided by the present invention is shown. Figure 4 As shown, step 21 above may include the following steps:

[0099] Step 41: In response to the user's selection of multiple preset scene types in the vehicle's interactive interface, determine the first driving scene;

[0100] In this step, users can select from several preset driving scenario types through the in-vehicle central control screen. These scenario types usually cover high-frequency, typical, or specific driving situations for users, such as daily commuting routes, long-distance driving on highways, urban congested roads, and driving at night in rainy weather.

[0101] At this point, the vehicle records the user's selection and identifies the selected scenario as the first driving scenario to be simulated and collected.

[0102] Step 42: Load the traffic condition data and driving area map data corresponding to the first driving scenario into the vehicle simulation game platform;

[0103] In this step, after determining the first driving scenario, based on the actual geographical location, route characteristics, and time attributes associated with the selected scenario, high-precision map information of the corresponding area is retrieved from the cloud database or local storage, and real-time or simulated traffic condition data is obtained simultaneously, including but not limited to: road topology, lane information, traffic light timing, dynamics of surrounding vehicles and pedestrians, congestion level, weather simulation, etc.

[0104] These traffic data and driving area map data are integrated and loaded into the built-in simulation game platform to create a virtual driving environment that is highly consistent with the user's real driving scenario, with immersion and behavioral responsiveness.

[0105] Step 43: Obtain the first driving behavior data corresponding to the user's driving operation in the simulation game platform.

[0106] In this step, the user performs interactive driving operations in the loaded simulation scenario. Through the steering wheel, pedals, central control interaction devices, etc., the user's driving control inputs are captured in real time, including: steering angle, acceleration and braking depth, gear shifting actions, etc., and the vehicle's operating status in the virtual environment is recorded simultaneously as the first driving behavior data.

[0107] The intelligent driving model training method provided in this invention determines a first driving scenario in response to a user's selection of multiple preset scenario types in the vehicle's interactive interface; loads traffic condition data and driving area map data corresponding to the first driving scenario into the vehicle's simulation game platform; and acquires first driving behavior data corresponding to the user's driving operations in the simulation game platform. This solution, by having the user interact to select a preset driving scenario and dynamically loading corresponding high-precision traffic and map data into the simulation platform, effectively constructs a training environment that closely resembles the user's real driving needs and typical road conditions. This ensures that the collected first driving behavior data has high scenario representativeness and personalized relevance, providing a highly targeted and reliable data foundation for subsequent personalized driving model training. Ultimately, this improves the intelligent driving system's adaptability to user habits and the reliability of its decisions in actual corresponding scenarios.

[0108] Based on the above embodiments, Figure 5 A flowchart of a fourth embodiment of the training method for the intelligent driving model provided by the present invention is shown. Figure 5 As shown, an overall example of this method may include the following steps:

[0109] Step 501: User logs into the game;

[0110] Step 502: Select a game scene;

[0111] Step 503: Load the real traffic scene / AEB trigger scene;

[0112] Step 504: Begin driving controls in the game;

[0113] Step 505: Collect driving behavior data; execute steps 506 and 509;

[0114] Step 506: Input into the behavior quantification module;

[0115] Step 507: Quantify driving behavior data into driving behavior parameters;

[0116] Step 508, Driving behavior gating module; Execute step 512;

[0117] Step 509: Input into the rule scoring module;

[0118] Step 510: Calculate the score;

[0119] Step 511: Rule-based scoring gating module; tiered reward and punishment mechanism, divided into: score less than A (forced safety learning, i.e., the first preset score); score greater than B (reward points, i.e., the second preset score); score between A and B (no measures).

[0120] Step 512: Pass? If yes, proceed to step 514; otherwise, proceed to step 513.

[0121] Step 513: Discard data;

[0122] Step 514: Mark as valid data;

[0123] Step 515: Input into the reinforcement learning trainer;

[0124] Step 516: Update the intelligent driving model algorithm;

[0125] Step 517: OTA deployment to the vehicle.

[0126] The training method for the intelligent driving model provided in this embodiment of the invention has the following technical advantages:

[0127] Technical advantages 1) Habit deep matching: The vehicle intelligent driving model algorithm is closer to the user's driving habits, making it easier for users to accept and use intelligent driving;

[0128] Technical advantage 2) Safe training environment: There are no real consequences for violations or accidents that occur in the game;

[0129] Technical advantage 3), cost reduction: It replaces the collection of millions of kilometers of real-world vehicle road test data, greatly reducing vehicle manufacturing costs.

[0130] Technical advantage 4) Close to real user scenarios: Recreates the user's main driving scenarios and the real traffic situation when AEB is triggered, enhancing the safety of users' daily driving scenarios.

[0131] Figure 6 A schematic diagram of an embodiment of the training device for the intelligent driving model provided by the present invention is shown. Figure 6 As shown, the device includes:

[0132] The acquisition module 61 is used to respond to the user's driving operation in the first driving scenario and acquire the first driving behavior data corresponding to the driving operation;

[0133] The determination module 62 is used to determine the target score corresponding to the first driving behavior data according to the score reward and punishment strategy. The score reward and punishment strategy is determined based on traffic rule constraints and / or preset route time.

[0134] Training module 63 is used to update the intelligent driving model deployed in the vehicle based on the first driving behavior data if the target score is greater than the first preset score.

[0135] In one or more embodiments, the training module 63 updates the intelligent driving model deployed in the vehicle based on the first driving behavior data, specifically for:

[0136] According to the filtering strategy, the first driving behavior data is filtered. The filtering strategy includes: at least one driving behavior parameter and a preset value range corresponding to each driving behavior parameter.

[0137] The intelligent driving model deployed in the vehicle is updated based on the filtered first driving behavior data.

[0138] In one or more embodiments, the training module 63 filters the first driving behavior data according to a filtering strategy, specifically for:

[0139] The first driving behavior data is quantified into multiple first driving behavior parameters;

[0140] According to the filtering strategy, driving behavior parameters that are not located in the corresponding preset value range among multiple first driving behavior parameters are filtered.

[0141] In one or more embodiments, the score-based reward and punishment strategy includes at least one of the following:

[0142] If driving behavior data indicates a violation of traffic rules, such as running a red light, the first point will be deducted.

[0143] If driving behavior data indicates an illegal lane change that violates traffic rules, a second point will be deducted.

[0144] If driving behavior data indicates a violation of traffic rule constraints regarding yielding to pedestrians, a third point will be deducted.

[0145] If driving behavior data indicates speeding in violation of traffic rules, the fourth point will be deducted based on the speeding percentage.

[0146] If the driving behavior data indicates that the actual route time is longer than the preset route time, the fifth point will be deducted based on the overtime ratio.

[0147] In one or more embodiments, the determining module 62 is further configured to:

[0148] If the target score is less than or equal to the first preset score, control the vehicle to perform the first operation, which is to instruct the user to learn traffic safety rules.

[0149] If the target score is greater than the second preset score, the first point is sent to the user's corresponding account. The first point can be used to redeem a preset object. The second preset score is greater than the first preset score.

[0150] In one or more embodiments, the acquisition module 61, in response to a user's driving operation in a first driving scenario, acquires first driving behavior data corresponding to the driving operation, specifically for:

[0151] In response to the user's selection of multiple preset scenario types in the vehicle's interactive interface, the first driving scenario is determined;

[0152] Load the traffic condition data and driving area map data corresponding to the first driving scenario into the vehicle simulation game platform;

[0153] Obtain the first driving behavior data corresponding to the user's driving operation in the simulation game platform.

[0154] In one or more embodiments, the first driving scenario is a real driving scenario;

[0155] Correspondingly, module 61 acquires the first driving behavior data corresponding to the driving operation, specifically for:

[0156] When AEB is detected to be triggered in the vehicle, the first driving behavior data corresponding to the driving operation is acquired.

[0157] It should be noted that the division of the various modules in the above device is merely a logical functional division. In actual implementation, they can be fully or partially integrated into a single physical element, or they can be physically separated. Furthermore, these modules can be implemented entirely in software through processing element calls, or entirely in hardware. Alternatively, some modules can be implemented through processing element calls in software, while others can be implemented in hardware. Moreover, these modules can be integrated together or implemented independently. The processing element here can be an integrated circuit with signal processing capabilities. During implementation, each step of the above method or each of the above modules can be completed through the integrated logic circuits in the hardware of the processor element or through software instructions.

[0158] As can be seen from the above, the training device for the intelligent driving model provided in this embodiment of the invention associates user driving behavior data with a score reward and punishment strategy based on traffic rules and traffic efficiency in real time. When the user's operation satisfies the target score greater than a preset threshold, the user's behavior data is used to update the intelligent driving model. This ensures that the data entering the model training meets the basic standards of safety and efficiency. Thus, in the process of continuous personalized optimization, it can gradually adapt to the user's driving habits and fundamentally avoid the model learning unsafe or inefficient driving modes. Ultimately, while improving driving comfort and acceptability, it also ensures the safety and compliance of the intelligent driving system.

[0159] Figure 7 A schematic diagram of an embodiment of the electronic device provided by the present invention is shown, such as... Figure 7 As shown, the electronic device may include: a processor 72, a communications interface 74, a memory 76, and a communications bus 78.

[0160] The processor 72, communication interface 74, and memory 76 communicate with each other via communication bus 78. Communication interface 74 is used to communicate with other network elements such as clients or other servers. The processor 72 executes program 70, specifically performing the relevant steps in the above method embodiments.

[0161] Specifically, program 70 may include program code, which includes computer-executable instructions.

[0162] Processor 72 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The electronic device includes one or more processors, which may be processors of the same type, such as one or more CPUs, or processors of different types, such as one or more CPUs and one or more ASICs.

[0163] Memory 76 is used to store program 70. Memory 76 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage device.

[0164] Specifically, program 70 can be called by processor 72 to cause the electronic device to perform the following operations:

[0165] In response to the user's driving operations in the first driving scenario, obtain the first driving behavior data corresponding to the driving operations;

[0166] Based on the score-based reward and punishment strategy, the target score corresponding to the first driving behavior data is determined. The score-based reward and punishment strategy is determined based on traffic rule constraints and / or preset route time.

[0167] If the target score is greater than the first preset score, the intelligent driving model deployed in the vehicle will be updated based on the first driving behavior data.

[0168] In one or more embodiments, updating the intelligent driving model deployed in the vehicle based on first driving behavior data includes:

[0169] According to the filtering strategy, the first driving behavior data is filtered. The filtering strategy includes: at least one driving behavior parameter and a preset value range corresponding to each driving behavior parameter.

[0170] The intelligent driving model deployed in the vehicle is updated based on the filtered first driving behavior data.

[0171] In one or more embodiments, the first driving behavior data is filtered according to a filtering strategy, including:

[0172] The first driving behavior data is quantified into multiple first driving behavior parameters;

[0173] According to the filtering strategy, driving behavior parameters that are not located in the corresponding preset value range among multiple first driving behavior parameters are filtered.

[0174] In one or more embodiments, the score-based reward and punishment strategy includes at least one of the following:

[0175] If driving behavior data indicates a violation of traffic rules, such as running a red light, the first point will be deducted.

[0176] If driving behavior data indicates an illegal lane change that violates traffic rules, a second point will be deducted.

[0177] If driving behavior data indicates a violation of traffic rule constraints regarding yielding to pedestrians, a third point will be deducted.

[0178] If driving behavior data indicates speeding in violation of traffic rules, the fourth point will be deducted based on the speeding percentage.

[0179] If the driving behavior data indicates that the actual route time is longer than the preset route time, the fifth point will be deducted based on the overtime ratio.

[0180] In one or more embodiments, the following is also performed:

[0181] If the target score is less than or equal to the first preset score, control the vehicle to perform the first operation, which is to instruct the user to learn traffic safety rules.

[0182] If the target score is greater than the second preset score, the first point is sent to the user's corresponding account. The first point can be used to redeem a preset object. The second preset score is greater than the first preset score.

[0183] In one or more embodiments, in response to a user's driving operation in a first driving scenario, first driving behavior data corresponding to the driving operation is obtained, including:

[0184] In response to the user's selection of multiple preset scenario types in the vehicle's interactive interface, the first driving scenario is determined;

[0185] Load the traffic condition data and driving area map data corresponding to the first driving scenario into the vehicle simulation game platform;

[0186] Obtain the first driving behavior data corresponding to the user's driving operation in the simulation game platform.

[0187] In one or more embodiments, the first driving scenario is a real driving scenario;

[0188] Accordingly, the first driving behavior data corresponding to the driving operation is obtained, including:

[0189] When AEB is detected to be triggered in the vehicle, the first driving behavior data corresponding to the driving operation is acquired.

[0190] As can be seen from the above, the electronic device provided in this embodiment of the invention can associate user driving behavior data with a score-based reward and punishment strategy based on traffic rules and traffic efficiency in real time. When the user's operation satisfies the target score greater than a preset threshold, the user's behavior data is used to update the intelligent driving model. This ensures that the data entering the model training meets the basic standards of safety and efficiency. Thus, in the process of continuous personalized optimization, it can gradually adapt to the user's driving habits and fundamentally avoid the model learning unsafe or inefficient driving modes. Ultimately, while improving driving comfort and acceptability, it ensures the safety and compliance of the intelligent driving system.

[0191] This invention provides a computer-readable storage medium storing at least one executable instruction. When the executable instruction is executed on a training device / electronic device for an intelligent driving model, it causes the training device / electronic device for the intelligent driving model to perform the training method for the intelligent driving model described in the above embodiments.

[0192] Specifically, the executable instructions can be used to cause the training device / electronic device of the intelligent driving model to perform the following operations:

[0193] In response to the user's driving operations in the first driving scenario, obtain the first driving behavior data corresponding to the driving operations;

[0194] Based on the score-based reward and punishment strategy, the target score corresponding to the first driving behavior data is determined. The score-based reward and punishment strategy is determined based on traffic rule constraints and / or preset route time.

[0195] If the target score is greater than the first preset score, the intelligent driving model deployed in the vehicle will be updated based on the first driving behavior data.

[0196] In one or more embodiments, updating the intelligent driving model deployed in the vehicle based on first driving behavior data includes:

[0197] According to the filtering strategy, the first driving behavior data is filtered. The filtering strategy includes: at least one driving behavior parameter and a preset value range corresponding to each driving behavior parameter.

[0198] The intelligent driving model deployed in the vehicle is updated based on the filtered first driving behavior data.

[0199] In one or more embodiments, the first driving behavior data is filtered according to a filtering strategy, including:

[0200] The first driving behavior data is quantified into multiple first driving behavior parameters;

[0201] According to the filtering strategy, driving behavior parameters that are not located in the corresponding preset value range among multiple first driving behavior parameters are filtered.

[0202] In one or more embodiments, the score-based reward and punishment strategy includes at least one of the following:

[0203] If driving behavior data indicates a violation of traffic rules, such as running a red light, the first point will be deducted.

[0204] If driving behavior data indicates an illegal lane change that violates traffic rules, a second point will be deducted.

[0205] If driving behavior data indicates a violation of traffic rule constraints regarding yielding to pedestrians, a third point will be deducted.

[0206] If driving behavior data indicates speeding in violation of traffic rules, the fourth point will be deducted based on the speeding percentage.

[0207] If the driving behavior data indicates that the actual route time is longer than the preset route time, the fifth point will be deducted based on the overtime ratio.

[0208] In one or more embodiments, the following is also performed:

[0209] If the target score is less than or equal to the first preset score, control the vehicle to perform the first operation, which is to instruct the user to learn traffic safety rules.

[0210] If the target score is greater than the second preset score, the first point is sent to the user's corresponding account. The first point can be used to redeem a preset object. The second preset score is greater than the first preset score.

[0211] In one or more embodiments, in response to a user's driving operation in a first driving scenario, first driving behavior data corresponding to the driving operation is obtained, including:

[0212] In response to the user's selection of multiple preset scenario types in the vehicle's interactive interface, the first driving scenario is determined;

[0213] Load the traffic condition data and driving area map data corresponding to the first driving scenario into the vehicle simulation game platform;

[0214] Obtain the first driving behavior data corresponding to the user's driving operation in the simulation game platform.

[0215] In one or more embodiments, the first driving scenario is a real driving scenario;

[0216] Accordingly, the first driving behavior data corresponding to the driving operation is obtained, including:

[0217] When AEB is detected to be triggered in the vehicle, the first driving behavior data corresponding to the driving operation is acquired.

[0218] As can be seen from the above, the training device / electronic device for the intelligent driving model provided in this embodiment of the invention can associate user driving behavior data with a score reward and punishment strategy based on traffic rules and traffic efficiency in real time. When the user's operation satisfies the target score greater than a preset threshold, the user's behavior data is used to update the intelligent driving model. This ensures that the data entering the model training meets the basic standards of safety and efficiency. Thus, in the process of continuous personalized optimization, it can gradually adapt to the user's driving habits and fundamentally avoid the model learning unsafe or inefficient driving modes. Ultimately, while improving driving comfort and acceptability, it ensures the safety and compliance of the intelligent driving system.

[0219] This invention provides a computer program product, including a computer program that, when executed by a processor, implements the above-described training method for an intelligent driving model.

[0220] Its implementation principle and technical effects are as disclosed above.

[0221] The description of the various embodiments above tends to emphasize the differences between the various embodiments. The similarities or similarities between them can be referred to, and for the sake of brevity, they will not be repeated here.

[0222] The methods disclosed in the various method embodiments provided by this invention can be arbitrarily combined without conflict to obtain new method embodiments.

[0223] The features disclosed in the various product embodiments provided by this invention can be arbitrarily combined without conflict to obtain new product embodiments.

[0224] The features disclosed in the various method or device embodiments provided by the present invention can be arbitrarily combined without conflict to obtain new method or device embodiments.

[0225] It should be noted that the aforementioned computer-readable storage media can be ROM, Programmable Read-Only Memory (PROM), Erasable Programmable Read-Only Memory (EPROM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Ferromagnetic Random Access Memory (FRAM), Flash Memory, Magnetic Surface Memory, Optical Disc, or Compact Disc Read-Only Memory (CD-ROM), etc. It can also be various vehicles that include one or any combination of the above-mentioned storage media.

[0226] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0227] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0228] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware nodes. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, vehicle terminal, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0229] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatuses, devices, and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0230] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0231] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The algorithms or displays provided herein for the functions specified in the boxes or boxes are not inherently related to any particular computer, virtual system, or other device. Furthermore, the embodiments of this invention are not directed to any particular programming language.

[0232] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be construed as limiting the order of execution.

Claims

1. A training method for an intelligent driving model, characterized in that, The method includes: In response to the user's driving operation in the first driving scenario, the first driving behavior data corresponding to the driving operation is obtained; According to the score reward and punishment strategy, the target score corresponding to the first driving behavior data is determined, and the score reward and punishment strategy is determined based on traffic rule constraints and / or preset route time; If the target score is greater than the first preset score, the intelligent driving model deployed in the vehicle is updated based on the first driving behavior data.

2. The method according to claim 1, characterized in that, The step of updating the intelligent driving model deployed in the vehicle based on the first driving behavior data includes: According to the filtering strategy, the first driving behavior data is filtered, and the filtering strategy includes: at least one driving behavior parameter and a preset value range corresponding to each driving behavior parameter. The intelligent driving model deployed in the vehicle is updated based on the filtered first driving behavior data.

3. The method according to claim 2, characterized in that, The step of filtering the first driving behavior data according to the filtering strategy includes: The first driving behavior data is quantified into multiple first driving behavior parameters; According to the filtering strategy, driving behavior parameters that are not located in the corresponding preset value range among the plurality of first driving behavior parameters are filtered.

4. The method according to any one of claims 1-3, characterized in that, The score-based reward and punishment strategy includes at least one of the following: If the driving behavior data indicates a violation of the aforementioned traffic rules, specifically running a red light, the first point will be deducted. If the driving behavior data indicates an illegal lane change that violates the aforementioned traffic rule constraints, a second point will be deducted; If driving behavior data indicates a violation of the traffic rule constraint of yielding to pedestrians, a third point will be deducted; If driving behavior data indicates speeding in violation of the aforementioned traffic rules, a fourth point will be deducted based on the speeding percentage. If the driving behavior data indicates that the actual route time is longer than the preset route time, a fifth score will be deducted based on the overtime ratio.

5. The method according to any one of claims 1-3, characterized in that, The method further includes: If the target score is less than or equal to the first preset score, the vehicle is controlled to perform a first operation, which is to instruct the user to learn traffic safety rules. If the target score is greater than the second preset score, the first points are sent to the user's corresponding account. The first points are used to redeem a preset object. The second preset score is greater than the first preset score.

6. The method according to any one of claims 1-3, characterized in that, The step of responding to a user's driving operation in the first driving scenario and acquiring the first driving behavior data corresponding to the driving operation includes: In response to the user's selection of multiple preset scene types in the vehicle's interactive interface, the first driving scene is determined; Load the traffic condition data and driving area map data corresponding to the first driving scenario into the vehicle's simulation game platform; Obtain the first driving behavior data corresponding to the driving operation performed by the user in the simulation game platform.

7. The method according to any one of claims 1-3, characterized in that, The first driving scenario is a real driving scenario; Accordingly, obtaining the first driving behavior data corresponding to the driving operation includes: When AEB is detected to be triggered in the vehicle, first driving behavior data corresponding to the driving operation is acquired.

8. A training device for an intelligent driving model, characterized in that, The device includes: The acquisition module is used to acquire first driving behavior data of the driving operation in response to the user's driving operation in the first driving scenario; The determination module is used to determine the target score corresponding to the first driving behavior data according to the score reward and punishment strategy, wherein the score reward and punishment strategy is determined based on traffic rule constraints and / or preset path time. The training module is used to update the intelligent driving model deployed in the vehicle based on the first driving behavior data if the target score is greater than the first preset score.

9. An electronic device, characterized in that, include: The processor, memory, communication interface, and communication bus are provided, wherein the processor, memory, and communication interface communicate with each other via the communication bus. The memory is used to store at least one executable instruction that causes the processor to perform the operation of the training method for the intelligent driving model as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The storage medium stores at least one executable instruction; When the executable instructions are executed on the training device / electronic device of the intelligent driving model, the training device / electronic device of the intelligent driving model performs the operation of the training method of the intelligent driving model as described in any one of claims 1-7.

Images