An automatic driving decision planning model optimization method, device and equipment

Abstract

The embodiment of the specification discloses an optimization method of an automatic driving decision planning model. The method determines the related traffic data of a traffic scene event involving a first vehicle by analyzing the traffic data reported by a target intelligent roadside device, so as to process the related traffic data by using a first automatic driving decision planning model, to obtain a more accurate first automatic driving decision result. Then, the target traffic data in a preset area for optimizing and verifying a second automatic driving decision planning model is determined from the related traffic data of the traffic scene event, so as to reduce the collection time of the target traffic data. Finally, the first automatic driving decision result and the second automatic driving decision planning model carried by a second vehicle are used for optimization, thereby effectively improving the optimization efficiency and accuracy of the second automatic driving decision planning model.

Description

Technical Field

[0001] This application relates to the field of computer data processing technology, and in particular to an optimization method, apparatus and device for an autonomous driving decision planning model. Background Technology

[0002] With the development of intelligent driving technology, intelligent driving vehicles can generate driving strategies using iteratively optimized single-vehicle autonomous driving decision-making and planning models during operation, without relying entirely on cloud control platforms. Currently, intelligent driving vehicles typically use environmental data collected by sensors within their perception range to generate driving decisions. This environmental data is also saved so that after updating and optimizing the single-vehicle autonomous driving decision-making and planning model based on the generated driving decisions, new driving strategies can be generated using the saved environmental data to test the iteratively optimized model. This involves offline model-in-the-loop testing and iterative optimization of the single-vehicle autonomous driving decision-making and planning model using vehicle data feedback. However, iterative optimization of the single-vehicle autonomous driving decision-making and planning model using vehicle data feedback requires collecting environmental data from numerous identical scenarios, which is time-consuming and hinders rapid updates and iterations of the model. Furthermore, the environmental data used for iterative optimization of driving strategies—collected by the sensors within the vehicle's perception range—has limitations. Therefore, the driving strategies generated by the iteratively optimized autonomous driving decision-making and planning model are only locally optimal solutions within the vehicle's perception range, rather than globally optimal solutions within a larger environmental scope.

[0003] Therefore, how to provide a method for optimizing autonomous driving decision-making and planning models so that these models can be rapidly iterated and optimized, and at the same time make the driving strategies generated by these models feasible or optimal in a wider range of environments, has become an urgent technical problem to be solved. Summary of the Invention

[0004] This specification provides an optimization method, apparatus, and device for an autonomous driving decision-making and planning model to address the local optima problem in existing model optimization methods.

[0005] To solve the above-mentioned technical problems, the embodiments in this specification are implemented as follows:

[0006] This specification provides an optimization method for an autonomous driving decision-making and planning model, which may include:

[0007] The intelligent connected cloud control platform acquires relevant traffic data related to traffic scenarios involving the first vehicle; the relevant traffic data for the traffic scenarios is determined by analyzing the traffic data reported by the target intelligent roadside equipment.

[0008] Using the first autonomous driving decision planning model, the relevant traffic data of the traffic scene event is processed to obtain the first autonomous driving decision result for the first vehicle in the traffic scene event;

[0009] From the relevant traffic data of the traffic scene event, target traffic data within a preset area is determined; the preset area is the area determined based on the vehicle perception range of the second vehicle, with the first vehicle as the perception center point.

[0010] The first autonomous driving decision result and the target traffic data are sent to the second vehicle; the second vehicle is used to optimize the second autonomous driving decision planning model mounted on the second vehicle using the first autonomous driving decision result and the target traffic data.

[0011] This specification provides an optimization method for an autonomous driving decision-making and planning model, which may include:

[0012] The target vehicle acquires training data; the training data includes: a first autonomous driving decision result and target traffic data; the first autonomous driving decision result is an autonomous driving decision result obtained by processing relevant traffic data of traffic scene events using a first autonomous driving decision planning model; the relevant traffic data of traffic scene events is determined by analyzing traffic data reported by the target intelligent roadside device; the target traffic data is relevant traffic data within a preset area determined from the relevant traffic data of the traffic scene events.

[0013] The training data is used to train the second autonomous driving decision planning model at the target vehicle, resulting in the trained second autonomous driving decision planning model.

[0014] This specification provides an optimization device for an autonomous driving decision-making and planning model, which may include:

[0015] The relevant traffic data acquisition module is used by the intelligent connected cloud control platform to acquire relevant traffic data of traffic scene events involving the first vehicle; the relevant traffic data of the traffic scene events is determined by analyzing the traffic data reported by the target intelligent roadside equipment.

[0016] The autonomous driving decision result generation module is used to process relevant traffic data of the traffic scene event using the first autonomous driving decision planning model to obtain the first autonomous driving decision result for the first vehicle in the traffic scene event.

[0017] The target traffic data determination module is used to determine target traffic data within a preset area from the relevant traffic data of the traffic scene event; the preset area is an area determined based on the vehicle perception range of the second vehicle, with the first vehicle as the perception center point.

[0018] The sending module is used to send the first autonomous driving decision result and the target traffic data to the second vehicle; the second vehicle is used to optimize the second autonomous driving decision planning model mounted on the second vehicle using the first autonomous driving decision result and the target traffic data.

[0019] This specification provides an optimization device for an autonomous driving decision-making and planning model, which may include:

[0020] The training data acquisition module is used to acquire training data for the target vehicle; the training data includes: a first autonomous driving decision result and target traffic data; the first autonomous driving result is the result generated by the first autonomous driving decision planning model based on relevant traffic data of traffic scene events.

[0021] The training module is used to train the second autonomous driving decision-making and planning model using the training data to obtain the trained second autonomous driving decision-making and planning model.

[0022] This specification provides an optimization device for an autonomous driving decision-making and planning model, comprising:

[0023] At least one processor; and,

[0024] A memory communicatively connected to the at least one processor; wherein,

[0025] The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to:

[0026] Acquire relevant traffic data related to the traffic scene event involving the first vehicle; the relevant traffic data for the traffic scene event is determined by analyzing the traffic data reported by the target intelligent roadside equipment;

[0027] Using the first autonomous driving decision planning model, the relevant traffic data of the traffic scene event is processed to obtain the first autonomous driving decision result for the first vehicle in the traffic scene event;

[0028] From the relevant traffic data of the traffic scene event, target traffic data within a preset area is determined; the preset area is the area determined based on the vehicle perception range of the second vehicle, with the first vehicle as the perception center point.

[0029] The first autonomous driving decision result and the target traffic data are sent to the second vehicle; the second vehicle is used to optimize the second autonomous driving decision planning model mounted on the second vehicle using the first autonomous driving decision result and the target traffic data.

[0030] At least one embodiment in this specification can achieve the following beneficial effects:

[0031] By analyzing the traffic data reported by the target intelligent roadside device, relevant traffic data for the traffic scenarios involving the first vehicle is determined. This allows the first autonomous driving decision-making and planning model to process the more comprehensive relevant traffic data, resulting in a more accurate first autonomous driving decision for the first vehicle. Then, from the relevant traffic data of the traffic scenarios, target traffic data is determined to optimize and validate the second autonomous driving decision-making and planning model for the second vehicle, reducing the target traffic data collection time. Finally, the first autonomous driving decision and the target traffic data are considered as traffic data collected by the sensors on the second vehicle. Because the relevant traffic data for the traffic scenarios involving the first vehicle, determined by analyzing the traffic data reported by the target intelligent roadside device, has good global coverage, the first autonomous driving decision is more accurate. Furthermore, by using the relevant traffic data of the traffic scenarios to determine the target traffic data, the second autonomous driving decision-making and planning model on the second vehicle can be optimized without requiring the second vehicle to collect the target traffic data, thus effectively improving the optimization efficiency and accuracy of the second autonomous driving decision-making and planning model. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a flowchart illustrating an optimization method for an autonomous driving decision-making and planning model provided in an embodiment of this specification.

[0034] Figure 2 This is a flowchart illustrating another optimization method for an autonomous driving decision-making and planning model provided in the embodiments of this specification;

[0035] Figure 3 The embodiments provided in this specification correspond to Figure 1 A schematic diagram of the structure of an optimization device for an autonomous driving decision-making and planning model;

[0036] Figure 4 The embodiments provided in this specification correspond to Figure 2 A schematic diagram of the structure of an optimization device for an autonomous driving decision-making and planning model;

[0037] Figure 5 The embodiments provided in this specification correspond to Figure 1 A schematic diagram of the structure of an optimization device for an autonomous driving decision-making and planning model.

[0038] Figure 6 The embodiments provided in this specification correspond to Figure 2 A schematic diagram of the structure of an optimization device for an autonomous driving decision-making and planning model. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of one or more embodiments of this specification clearer, the technical solutions of one or more embodiments of this specification will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of them. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of one or more embodiments of this specification.

[0040] The technical solutions provided in the various embodiments of this specification are described in detail below with reference to the accompanying drawings.

[0041] In existing technologies, the iterative optimization of autonomous driving decision-making and planning models typically utilizes environmental data collected by sensors within their perception range to generate driving decisions. This environmental data is then saved so that after updating and optimizing the single-vehicle intelligent driving decision-making and planning model based on the generated driving decisions, the saved environmental data is used to generate new driving strategies to verify the iteratively optimized model. This involves offline model-in-the-loop testing and iterative optimization of the single-vehicle intelligent driving decision-making and planning model using vehicle data feedback. However, this method requires collecting numerous environmental data points from the vehicle in the same scenario, which is time-consuming and hinders rapid updates and iterations of the model. Furthermore, the environmental data used for iterative optimization of driving strategies—collected by sensors within the vehicle's perception range—has limitations. Therefore, the driving strategies generated by the iteratively optimized autonomous driving decision-making and planning model are only locally optimal solutions within the vehicle's perception range, rather than globally optimal solutions within a larger environmental context.

[0042] To address the shortcomings of existing technologies, this solution provides the following embodiments:

[0043] Figure 1 This is a flowchart illustrating an optimization method for an autonomous driving decision-making and planning model provided in an embodiment of this specification. From a programming perspective, the entity executing the process can be a cloud control platform or application client used to optimize the autonomous driving decision-making and planning model.

[0044] like Figure 1 As shown, the process may include the following steps:

[0045] Step 102: The intelligent connected cloud control platform acquires relevant traffic data of the traffic scene event involving the first vehicle; the relevant traffic data of the traffic scene event is determined by analyzing the traffic data reported by the target intelligent roadside equipment.

[0046] In the embodiments of this specification, the first vehicle can be a vehicle that has experienced a traffic scene event, and the traffic scene event can be any one of overtaking in the left lane, overtaking in the right lane, exiting a ramp, changing lanes in the left lane, or changing lanes in the right lane.

[0047] In the embodiments of this specification, the relevant traffic data obtained for different traffic scenarios may be different. For example, if the traffic scenario event that occurs to the first vehicle is changing lanes to the right, the relevant traffic data obtained may include the vehicle operation data of the first vehicle, the operation data of the vehicles in front of and behind the first vehicle in the same lane, the distances between the vehicles in front and behind and the first vehicle, and the operation data of the vehicles within a preset distance from the first vehicle in the lane where the first vehicle changed lanes.

[0048] In the embodiments of this specification, the traffic data reported by the target intelligent roadside device can be the perception data of the perception objects within its perception range obtained by the target intelligent roadside device. The perception objects may include a first vehicle, and the perception data may include the perception position, perception speed, perception heading angle, and other data of the perception objects. The relevant traffic data of the first vehicle can be obtained by calculating based on the perception data within its perception range obtained by the target intelligent roadside device.

[0049] Step 104: Using the first autonomous driving decision planning model, process the relevant traffic data of the traffic scene event to obtain the first autonomous driving decision result for the first vehicle in the traffic scene event.

[0050] In the embodiments of this specification, the first autonomous driving decision planning model may be a decision planning model that has been deployed on a cloud platform, has been trained, and can be used to generate driving decision and control strategy information.

[0051] In the embodiments of this specification, since the relevant traffic data is determined by analyzing the traffic data reported by the target intelligent roadside device, the relevant traffic data is more global than the data collected by the first vehicle using its onboard sensors, so that the decision results obtained when using the first autonomous driving decision planning model can be more accurate.

[0052] In the embodiments of this specification, the first autonomous driving decision result can be used as the true value to train and optimize the second autonomous driving decision planning model that needs to be trained and optimized in this application.

[0053] Step 106: Determine the target traffic data within a preset area from the relevant traffic data of the traffic scene event; the preset area is the area determined based on the vehicle perception range of the second vehicle, with the first vehicle as the perception center point.

[0054] In the embodiments of this specification, the target traffic data within the preset area can be data obtained by the second vehicle within its perception range using its own sensors, assuming the second vehicle is located at the position of the first vehicle; the target traffic data can be used to train, optimize, and verify the second autonomous driving decision-making and planning model.

[0055] In the embodiments of this specification, the target traffic data can be determined from the relevant traffic data based on the perception range of the second vehicle, thereby avoiding the need to use the second vehicle to generate various different traffic scenario events and then obtain the sensor perception data in the traffic scenario events. This allows for the rapid and large-scale acquisition of target traffic data, which is then used to optimize the second autonomous driving decision-making and planning model.

[0056] Step 108: Send the first autonomous driving decision result and the target traffic data to the second vehicle; the second vehicle is used to optimize the second autonomous driving decision planning model mounted on the second vehicle using the first autonomous driving decision result and the target traffic data.

[0057] In the embodiments described in this specification, the second vehicle can generate a driving strategy based on the perception data acquired by its onboard sensors, and the second vehicle is equipped with a second autonomous driving decision-making and planning model.

[0058] In the embodiments of this specification, the first autonomous driving decision result can be used as the true value, and the target traffic data can be used as training data to train the second autonomous driving decision planning model.

[0059] It should be understood that the order of some steps in the methods described in one or more embodiments of this specification may be interchanged according to actual needs, or some steps may be omitted or deleted.

[0060] Figure 1 The method described herein analyzes traffic data reported by a target intelligent roadside device to determine relevant traffic data for traffic scenarios involving a first vehicle. This data is then processed using a first autonomous driving decision-making and planning model to obtain a first autonomous driving decision result for the first vehicle in the traffic scenario. Next, target traffic data within a preset area is determined from the relevant traffic data. Finally, the first autonomous driving decision result and the target traffic data are used to optimize a second autonomous driving decision-making and planning model mounted on a second vehicle. Because the relevant traffic data for the traffic scenarios involving the first vehicle, determined from the traffic data reported by the target intelligent roadside device, is more comprehensive, the first autonomous driving decision result obtained by processing the relevant traffic data using the first autonomous driving decision-making and planning model is more accurate. Furthermore, determining the target traffic data for optimizing and validating the second autonomous driving decision-making and planning model from the relevant traffic data effectively reduces data collection time, thereby significantly improving the optimization efficiency and accuracy of the second autonomous driving decision-making and planning model.

[0061] based on Figure 1 In addition to the method described herein, this specification also provides some specific implementation methods of this method, which will be described below.

[0062] In practical applications, there are countless vehicles on the road network at any given moment, but not all vehicle operation data can be used to train and optimize the second autonomous driving decision-making and planning model. Therefore, it is necessary to filter the data to select data related to traffic scene events, so as to obtain data that can be used to train the second autonomous driving decision-making and planning model.

[0063] Based on this, before the intelligent connected cloud control platform acquires the relevant traffic data of the traffic scene event involving the first vehicle, it may also include:

[0064] Acquire vehicle operation data reported by vehicles with data upload capabilities; the vehicle operation data includes vehicle driving parameters and vehicle location.

[0065] Determine whether the vehicle driving parameters meet the target recognition rules for the target traffic scene event, and obtain the determination result.

[0066] If the judgment result indicates that the vehicle driving parameters meet the target recognition rule, then the vehicle is identified as the first vehicle.

[0067] The target intelligent roadside device is determined based on the vehicle location carried by the first vehicle; the first vehicle is located within the effective sensing area of ​​the target intelligent roadside device during the time period of the target traffic scene event.

[0068] Obtain the specified traffic data reported by the target intelligent roadside device during the time period of the target traffic scenario event.

[0069] The intelligent connected cloud control platform acquires relevant traffic data related to traffic scenarios involving the first vehicle, which may specifically include:

[0070] The intelligent connected cloud control platform acquires the specified traffic data reported by the target intelligent roadside equipment during the time period of the target traffic scenario event.

[0071] The relevant traffic data of the traffic scene event involving the first vehicle is determined from the specified traffic data. The relevant traffic data of the traffic scene event may include: the vehicle operation data of the specified vehicle within a preset range from the first vehicle and the relative operation data of the specified vehicle within the preset range from the first vehicle and the first vehicle.

[0072] In the embodiments of this specification, a vehicle with data uploading capability can use its own data uploading module to transmit its own vehicle operation data to the vehicle bus. The vehicle operation data may include vehicle driving parameters, specifically including: operating data of various systems within the vehicle, such as engine, transmission, and braking system operating data; and may also include operating data such as vehicle speed, acceleration, and steering wheel angle. The vehicle operation data may also include vehicle position. In practical applications, the vehicle can use its own positioning module to locate its position in real time to ensure positioning accuracy.

[0073] In the embodiments of this specification, different traffic scene events correspond to different traffic scene event identification rules. When the vehicle operation data meets a certain specific target identification rule in the traffic scene event identification rules, it indicates that the vehicle has experienced a traffic scene event corresponding to the target identification rule that is met.

[0074] In the embodiments of this specification, the vehicle in which the target traffic scene event occurs can be identified as the first vehicle, so as to determine the target intelligent roadside equipment based on the location of the first vehicle, thereby obtaining the perception data of the target intelligent roadside equipment during the time period in which the target traffic scene event occurs in the first vehicle.

[0075] In the embodiments described in this specification, the specified traffic data may be the perception data of the target intelligent roadside device during the time period in which the first vehicle experiences the target traffic scenario event.

[0076] In practical applications, different second-stage autonomous driving decision-making and planning models can be trained for different traffic scenarios to improve the accuracy of the driving decisions output by the second-stage autonomous driving decision-making and planning models.

[0077] Based on this, before processing the relevant traffic data of the traffic scene event using the first autonomous driving decision-making and planning model to obtain the first autonomous driving decision result for the first vehicle in the traffic scene event, the process may further include:

[0078] Clustering is performed on the relevant traffic data of the traffic scene events to obtain the clustered traffic data of the traffic scene events.

[0079] The process of using a first autonomous driving decision-making and planning model to process relevant traffic data related to the traffic scenario event and obtain a first autonomous driving decision result for the first vehicle in the traffic scenario event may specifically include:

[0080] Using the first autonomous driving decision planning model, the clustered traffic data is processed to obtain the first autonomous driving decision result for the first vehicle in the traffic scenario event.

[0081] The step of determining target traffic data within a preset area from relevant traffic data of the traffic scene event may specifically include:

[0082] From the clustered traffic data, the first vehicle that caused the target traffic scene event is identified.

[0083] Using the location of the first vehicle as the perception center point, the perception data within the perception area defined by the perception range of the second vehicle is determined as the target traffic data.

[0084] In the embodiments of this specification, clustering methods such as K-means can be used to cluster related traffic data in the same scenario. Since the related traffic data after clustering are relatively similar, the clustered data can be more general and more representative of a specific traffic scenario, which can facilitate the subsequent training of the second autonomous driving decision planning model.

[0085] The above method analyzes the traffic data reported by the target intelligent roadside device to determine the relevant traffic data of the traffic scene events involving the first vehicle. This allows the first autonomous driving decision-making and planning model to process the relevant traffic data with a better global perspective, thereby obtaining a more accurate first autonomous driving decision result. Then, from the relevant traffic data of the traffic scene events, target traffic data within a preset area is determined, reducing the target traffic data collection time. Finally, the first autonomous driving decision result and the target traffic data are used to optimize the second autonomous driving decision-making and planning model mounted on the second vehicle, thereby effectively improving the optimization efficiency and accuracy of the second autonomous driving decision-making and planning model.

[0086] Figure 2 This is a flowchart illustrating another optimization method for an autonomous driving decision-making and planning model provided in the embodiments of this specification. From a program perspective, the entity executing the process can be the target vehicle or a cloud control platform used to optimize the autonomous driving decision-making and planning model.

[0087] like Figure 2 As shown, the process may include the following steps:

[0088] Step 202: The target vehicle acquires training data; the training data includes: a first autonomous driving decision result and target traffic data; the first autonomous driving decision result is an autonomous driving decision result obtained by processing relevant traffic data of traffic scene events using a first autonomous driving decision planning model; the relevant traffic data of traffic scene events is determined by analyzing the traffic data reported by the target intelligent roadside device; the target traffic data is relevant traffic data within a preset area determined from the relevant traffic data of the traffic scene events.

[0089] In the embodiments of this specification, the first autonomous driving decision planning model may be a decision planning model that has been deployed on a cloud platform, has been trained, and can be used to generate driving decision and control strategy information.

[0090] In the embodiments of this specification, the traffic data reported by the target intelligent roadside device can be the perception data of the sensing objects within its perception range obtained by the target intelligent roadside device. The sensing objects may include a first vehicle, and the perception data may include the sensing position, sensing speed, sensing heading angle, and other data of the sensing objects. The relevant traffic data of the first vehicle can be obtained by calculating based on the perception data within its perception range obtained by the target intelligent roadside device.

[0091] In the embodiments of this specification, since the relevant traffic data is determined by analyzing the traffic data reported by the target intelligent roadside equipment, the relevant traffic data is more comprehensive than the data collected by the vehicle using its onboard sensors. Therefore, when the first autonomous driving decision planning model is used for decision planning, the decision results obtained can be more accurate and more comprehensive.

[0092] Step 204: Use the training data to train the second autonomous driving decision planning model at the target vehicle to obtain the trained second autonomous driving decision planning model.

[0093] In the embodiments of this specification, the first autonomous driving decision result can be used as the true value, and the target traffic data can be used as training data to train the second autonomous driving decision planning model.

[0094] Figure 2 The method described above analyzes traffic data reported by the target intelligent roadside device to determine relevant traffic data for the traffic scenarios involving the first vehicle. This allows the first autonomous driving decision-making and planning model to process the relevant traffic data with a better global perspective, thereby obtaining a more accurate first autonomous driving decision result. Then, from the relevant traffic data of the traffic scenarios, target traffic data within a preset area is determined, reducing the target traffic data collection time. Finally, the first autonomous driving decision result and the target traffic data are used to optimize the second autonomous driving decision-making and planning model mounted on the second vehicle, thereby effectively improving the optimization efficiency and accuracy of the second autonomous driving decision-making and planning model.

[0095] In the embodiments of this specification, a specific process for training the second autonomous driving decision-making and planning model is also provided.

[0096] Specifically, training the second autonomous driving decision-making and planning model using the training data may include:

[0097] The second autonomous driving decision planning model is used to process the target traffic data to obtain the second autonomous driving decision result.

[0098] Based on the first autonomous driving decision result and the second autonomous driving decision result, the accuracy of the second autonomous driving decision planning model is determined.

[0099] If the accuracy rate is less than the preset accuracy rate, then the training samples are used to continue training the second autonomous driving decision planning model until the accuracy rate of the second autonomous driving decision planning model is greater than the preset accuracy rate value.

[0100] Based on the same idea, embodiments of this specification also provide apparatus corresponding to the above methods. Figure 3 The embodiments provided in this specification correspond to Figure 1 A schematic diagram of the structure of an optimization device for an autonomous driving planning model. (See attached diagram.) Figure 3 As shown, the device may include:

[0101] The relevant traffic data acquisition module 302 is used by the intelligent connected cloud control platform to acquire relevant traffic data of the traffic scene events involving the first vehicle; the relevant traffic data of the traffic scene events is determined by analyzing the traffic data reported by the target intelligent roadside equipment.

[0102] The autonomous driving decision result generation module 304 is used to process the relevant traffic data of the traffic scene event using the first autonomous driving decision planning model to obtain the first autonomous driving decision result for the first vehicle in the traffic scene event.

[0103] The target traffic data determination module 306 is used to determine target traffic data within a preset area from the relevant traffic data of the traffic scene event; the preset area is an area determined based on the vehicle perception range of the second vehicle, with the first vehicle as the perception center point.

[0104] The sending module 308 is used to send the first autonomous driving decision result and the target traffic data to the second vehicle; the second vehicle is used to optimize the second autonomous driving decision planning model mounted on the second vehicle using the first autonomous driving decision result and the target traffic data.

[0105] based on Figure 3 The embodiments of this specification also provide some specific implementation schemes of the method, which are described below.

[0106] Optional, Figure 3 The device may further include:

[0107] The vehicle data acquisition module is used to acquire vehicle operation data reported by vehicles with data upload function; the vehicle operation data includes vehicle driving parameters and vehicle location.

[0108] The judgment module is used to determine whether the vehicle driving parameters meet the target recognition rules of the target traffic scene event, and obtain the judgment result.

[0109] The first vehicle determination module is used to determine the vehicle as the first vehicle if the judgment result indicates that the vehicle's driving parameters meet the target classification rule.

[0110] The target intelligent roadside device determination module is used to determine the target intelligent roadside device based on the vehicle location carried by the first vehicle; the first vehicle is located within the effective perception area of ​​the target intelligent roadside device during the time period of the occurrence of the target traffic scene event.

[0111] A designated traffic data acquisition module is used to acquire designated traffic data reported by the target intelligent roadside device during the time period of the target traffic scenario event.

[0112] The relevant traffic data acquisition module 302 can be specifically used for:

[0113] The intelligent connected cloud control platform acquires the specified traffic data reported by the target intelligent roadside equipment during the time period of the target traffic scenario event.

[0114] The relevant traffic data of the traffic scene event involving the first vehicle is determined from the specified traffic data. The relevant traffic data of the traffic scene event includes: the vehicle operation data of the specified vehicle within a preset range from the first vehicle and the relative operation data of the specified vehicle within the preset range from the first vehicle and the first vehicle.

[0115] Optional, Figure 3 The device may further include:

[0116] The clustering module is used to perform clustering processing on the relevant traffic data of the traffic scene event to obtain the clustered traffic data of the traffic scene event.

[0117] The autonomous driving decision result generation module 304 can be specifically used for:

[0118] Using the first autonomous driving decision planning model, the clustered traffic data is processed to obtain the first autonomous driving decision result for the first vehicle in the traffic scenario event.

[0119] Optionally, the target traffic data determination module 306 can be specifically used for:

[0120] From the clustered traffic data, the first vehicle that caused the target traffic scene event is identified.

[0121] Using the location of the first vehicle as the perception center point, the perception data within the perception area defined by the perception range of the second vehicle is determined as the target traffic data.

[0122] Based on the same idea, embodiments of this specification also provide apparatus corresponding to the above methods. Figure 4 The embodiments provided in this specification correspond to Figure 2A schematic diagram of the structure of an optimization device for an autonomous driving planning model. (See attached diagram.) Figure 4 As shown, the device may include:

[0123] The training data acquisition module 402 is used to acquire training data for the target vehicle; the training data includes: a first autonomous driving decision result and target traffic data; the first autonomous driving result is the result generated by the first autonomous driving decision planning model based on relevant traffic data of traffic scene events.

[0124] The training module 404 is used to train the second autonomous driving decision planning model using the training data to obtain the trained second autonomous driving decision planning model.

[0125] based on Figure 4 The embodiments of this specification also provide some specific implementation schemes of the method, which are described below.

[0126] Optionally, the training module 404 can be specifically used for:

[0127] The second autonomous driving decision planning model is used to process the target traffic data to obtain the second autonomous driving decision result.

[0128] Based on the first autonomous driving decision result and the second autonomous driving decision result, the accuracy of the second autonomous driving decision planning model is determined.

[0129] If the accuracy rate is less than the preset accuracy rate, then the training samples are used to continue training the second autonomous driving decision planning model until the accuracy rate of the second autonomous driving decision planning model is greater than the preset accuracy rate value.

[0130] Based on the same idea, this specification also provides devices corresponding to the above methods in its embodiments.

[0131] Figure 5 The embodiments provided in this specification correspond to Figure 1 A schematic diagram of the structure of an optimization device for an autonomous driving decision-making and planning model. (See diagram below.) Figure 5 As shown, device 500 may include:

[0132] At least one processor 510; and,

[0133] Memory 530 communicatively connected to the at least one processor; wherein,

[0134] The memory 530 stores instructions 520 that can be executed by the at least one processor 510, the instructions being executed by the at least one processor 510 to enable the at least one processor 510 to:

[0135] Acquire relevant traffic data for the traffic scene event involving the first vehicle; the relevant traffic data for the traffic scene event is determined by analyzing the traffic data reported by the target intelligent roadside equipment.

[0136] Using the first autonomous driving decision planning model, the relevant traffic data of the traffic scene event is processed to obtain the first autonomous driving decision result for the first vehicle in the traffic scene event.

[0137] From the relevant traffic data of the traffic scene event, target traffic data within a preset area is determined; the preset area is the area determined with the first vehicle as the perception center point based on the vehicle perception range of the second vehicle.

[0138] The first autonomous driving decision result and the target traffic data are sent to the second vehicle; the second vehicle is used to optimize the second autonomous driving decision planning model mounted on the second vehicle using the first autonomous driving decision result and the target traffic data.

[0139] Figure 6 The embodiments provided in this specification correspond to Figure 2 A schematic diagram of the structure of an optimization device for an autonomous driving decision-making and planning model. (See diagram below.) Figure 6 As shown, device 600 may include:

[0140] At least one processor 610; and,

[0141] Memory 630 communicatively connected to the at least one processor; wherein,

[0142] The memory 630 stores instructions 620 that can be executed by the at least one processor 610, the instructions being executed by the at least one processor 610 to enable the at least one processor 610 to:

[0143] Acquire training data; the training data includes: a first autonomous driving decision result and target traffic data; the first autonomous driving decision result is an autonomous driving decision result obtained by processing relevant traffic data of traffic scene events using a first autonomous driving decision planning model; the relevant traffic data of traffic scene events is determined by analyzing traffic data reported by target intelligent roadside devices; the target traffic data is relevant traffic data within a preset area determined from the relevant traffic data of traffic scene events.

[0144] The training data is used to train the second autonomous driving decision planning model at the target vehicle, resulting in the trained second autonomous driving decision planning model.

[0145] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on its differences from other embodiments. In particular, for... Figure 5 , Figure 6 As for the optimization device of the autonomous driving decision-making and planning model shown, since it is basically similar to the method embodiment, the description is relatively simple, and relevant parts can be referred to in the description of the method embodiment.

[0146] In the 1990s, improvements to a technology could be clearly distinguished as either hardware improvements (e.g., improvements to the circuit structure of diodes, transistors, switches, etc.) or software improvements (improvements to the methodology). However, with technological advancements, many methodological improvements today can be considered direct improvements to the hardware circuit structure. Designers almost always obtain the corresponding hardware circuit structure by programming the improved methodology into the hardware circuit. Therefore, it cannot be said that a methodological improvement cannot be implemented using hardware physical modules. For example, a Programmable Logic Device (PLD) (such as a Field Programmable Gate Array (FPGA)) is such an integrated circuit whose logic function is determined by the user programming the device. Designers can program and "integrate" a digital system onto a PLD themselves, without needing chip manufacturers to design and manufacture dedicated integrated circuit chips. Furthermore, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software. Similar to the software compiler used in program development, the original code before compilation must be written in a specific programming language, called a Hardware Description Language (HDL). There are many HDLs, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, and RHDL (Ruby Hardware Description Language). Currently, the most commonly used are VHDL (Very-High-Speed ​​Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should understand that by simply performing some logic programming on the method flow using one of these hardware description languages ​​and programming it into an integrated circuit, the hardware circuit implementing the logical method flow can be easily obtained.

[0147] The controller can be implemented in any suitable manner. For example, it can take the form of a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code form, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included therein for implementing various functions can also be considered as structures within the hardware component. Alternatively, the means for implementing various functions can be considered as both software modules implementing the method and structures within the hardware component.

[0148] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0149] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0150] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0151] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), 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 illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0152] 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.

[0153] 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 steps of the function specified in one or more boxes.

[0154] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0155] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0156] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0157] It should also be noted that 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. Without further limitation, 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 said element.

[0158] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0159] This application can be described in the general context of computer-executable instructions, such as program modules, that are executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0160] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. An optimization method for an autonomous driving decision-making and planning model, characterized in that, The method includes: The intelligent connected cloud control platform acquires relevant traffic data related to traffic scenarios involving the first vehicle; the relevant traffic data for the traffic scenarios is determined by analyzing the traffic data reported by the target intelligent roadside equipment. Using the first autonomous driving decision planning model, the relevant traffic data of the traffic scene event is processed to obtain the first autonomous driving decision result for the first vehicle in the traffic scene event; From the relevant traffic data of the traffic scene event, target traffic data within a preset area is determined; the preset area is the area determined based on the vehicle perception range of the second vehicle, with the first vehicle as the perception center point. The first autonomous driving decision result and the target traffic data are sent to the second vehicle; the second vehicle is used to optimize the second autonomous driving decision planning model mounted on the second vehicle using the first autonomous driving decision result and the target traffic data.

2. The method according to claim 1, characterized in that, Before the intelligent connected cloud control platform acquires relevant traffic data related to the traffic scene event involving the first vehicle, it also includes: Acquire vehicle operation data reported by vehicles with data upload capabilities; the vehicle operation data includes vehicle driving parameters and vehicle location; Determine whether the vehicle driving parameters meet the target recognition rules for the target traffic scene event, and obtain the determination result; If the judgment result indicates that the vehicle driving parameters meet the target classification rule, then the vehicle is identified as the first vehicle; The target intelligent roadside device is determined based on the vehicle location carried by the first vehicle; the first vehicle is located within the effective sensing area of ​​the target intelligent roadside device during the time period of the occurrence of the target traffic scene event. Acquire the specified traffic data reported by the target intelligent roadside device during the time period of the occurrence of the target traffic scenario event; The intelligent connected cloud control platform acquires relevant traffic data related to the traffic scene events involving the first vehicle, specifically including: The intelligent connected cloud control platform acquires the specified traffic data reported by the target intelligent roadside equipment during the time period of the occurrence of the target traffic scenario event; The relevant traffic data of the traffic scene event involving the first vehicle is determined from the specified traffic data. The relevant traffic data of the traffic scene event includes: the vehicle operation data of the specified vehicle within a preset range from the first vehicle and the relative operation data of the specified vehicle within the preset range from the first vehicle and the first vehicle.

3. The method according to claim 1, characterized in that, Before processing the relevant traffic data of the traffic scene event using the first autonomous driving decision-making and planning model to obtain the first autonomous driving decision result for the first vehicle in the traffic scene event, the method further includes: Clustering is performed on the relevant traffic data of the traffic scene events to obtain the clustered traffic data of the traffic scene events; The process of using a first autonomous driving decision-making and planning model to process relevant traffic data related to the traffic scenario event and obtain a first autonomous driving decision result for the first vehicle in the traffic scenario event specifically includes: Using the first autonomous driving decision planning model, the clustered traffic data is processed to obtain the first autonomous driving decision result for the first vehicle in the traffic scenario event.

4. The method according to claim 3, characterized in that, The step of determining target traffic data within a preset area from relevant traffic data of the traffic scene event specifically includes: From the clustered traffic data, determine the first vehicle in which the target traffic scenario event occurred; Using the location of the first vehicle as the perception center point, the perception data within the perception area defined by the perception range of the second vehicle is determined as the target traffic data.

5. The method according to claim 1, characterized in that, The traffic scenario events include any one of the following: overtaking in the left lane, overtaking in the right lane, exiting a ramp, changing lanes in the left lane, and changing lanes in the right lane.

6. An optimization method for an autonomous driving decision-making and planning model, characterized in that, The method includes: The target vehicle acquires training data; the training data includes: a first autonomous driving decision result and target traffic data; the first autonomous driving decision result is an autonomous driving decision result obtained by processing relevant traffic data of traffic scene events using a first autonomous driving decision planning model; the relevant traffic data of traffic scene events is determined by analyzing traffic data reported by the target intelligent roadside device; the target traffic data is relevant traffic data within a preset area determined from the relevant traffic data of the traffic scene events. The training data is used to train the second autonomous driving decision planning model for the second vehicle in the target traffic data, resulting in the trained second autonomous driving decision planning model.

7. The method according to claim 6, characterized in that, The process of training the second autonomous driving decision-making and planning model using the training data specifically includes: The second autonomous driving decision-making and planning model is used to process the target traffic data to obtain the second autonomous driving decision result; Based on the first autonomous driving decision result and the second autonomous driving decision result, determine the accuracy of the second autonomous driving decision planning model; If the accuracy rate is less than the preset accuracy rate, then the training data is used to continue training the second autonomous driving decision planning model until the accuracy rate of the second autonomous driving decision planning model is greater than the preset accuracy rate value.

8. An optimization device for an autonomous driving planning model, characterized in that, The device includes: The relevant traffic data acquisition module is used by the intelligent connected cloud control platform to acquire relevant traffic data of traffic scene events involving the first vehicle; the relevant traffic data of the traffic scene events is determined by analyzing the traffic data reported by the target intelligent roadside equipment. The autonomous driving decision result generation module is used to process relevant traffic data of the traffic scene event using the first autonomous driving decision planning model to obtain the first autonomous driving decision result for the first vehicle in the traffic scene event. The target traffic data determination module is used to determine target traffic data within a preset area from the relevant traffic data of the traffic scene event; the preset area is an area determined based on the vehicle perception range of the second vehicle, with the first vehicle as the perception center point. The sending module is used to send the first autonomous driving decision result and the target traffic data to the second vehicle; the second vehicle is used to optimize the second autonomous driving decision planning model mounted on the second vehicle using the first autonomous driving decision result and the target traffic data.

9. An optimization device for an autonomous driving decision-making and planning model, characterized in that, The device includes: The training data acquisition module is used for the target vehicle to acquire training data. The training data includes: a first autonomous driving decision result and target traffic data. The first autonomous driving decision result is generated by the first autonomous driving decision planning model based on relevant traffic data of traffic scene events. The relevant traffic data of traffic scene events is determined by analyzing the traffic data reported by the target intelligent roadside device. The target traffic data is relevant traffic data within a preset area determined from the relevant traffic data of the traffic scene events. The training module is used to train the second autonomous driving decision-making and planning model at the second vehicle in the target traffic data using the training data, so as to obtain the trained second autonomous driving decision-making and planning model.

10. An optimization device for an autonomous driving decision-making and planning model, characterized in that, The device is located at the intelligent connected cloud control platform, and the device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to: Acquire relevant traffic data related to the traffic scene event involving the first vehicle; the relevant traffic data for the traffic scene event is determined by analyzing the traffic data reported by the target intelligent roadside equipment; Using the first autonomous driving decision planning model, the relevant traffic data of the traffic scene event is processed to obtain the first autonomous driving decision result for the first vehicle in the traffic scene event; From the relevant traffic data of the traffic scene event, target traffic data within a preset area is determined; the preset area is the area determined based on the vehicle perception range of the second vehicle, with the first vehicle as the perception center point. The first autonomous driving decision result and the target traffic data are sent to the second vehicle; the second vehicle is used to optimize the second autonomous driving decision planning model mounted on the second vehicle using the first autonomous driving decision result and the target traffic data.

Images