A method, apparatus, vehicle, equipment, and medium for iterative driving algorithm.

Abstract

This invention provides a method, apparatus, vehicle, device, and medium for iterative vehicle driving algorithm development. It utilizes a single software program to achieve rapid adaptation to different sensor types and communication methods across various vehicle models, and efficiently leverages sensor data to drive iterative updates of the driving algorithm for intelligent driving vehicles. The method includes: acquiring raw sensor data required by the vehicle based on a sensor configuration file; converting the acquired raw sensor data into a unified format of abstract sensor data and publishing it; sending the required abstract sensor data to the cloud; receiving the updated driving algorithm from the cloud based on the abstract sensor data, and performing iterative updates to the driving algorithm; the sensor configuration file is pre-formed by adding or removing sensor instances from historical sensor configuration files already applied to previous vehicle models based on the sensor configuration requirements of the newly developed vehicle model.

Description

Technical Field

[0001] This invention relates to the field of intelligent driving vehicle control, specifically to a method, apparatus, vehicle, equipment, and medium for iterative vehicle driving algorithm. Background Technology

[0002] With the rapid development of intelligent driving vehicles, the perception, decision-making, and execution technologies of intelligent driving have been greatly improved. Because multi-sensor fusion solutions are safer than pure vision solutions, they are gradually becoming the mainstream technology for intelligent driving. How to quickly and stably adapt to various automotive sensors, efficiently utilize sensor data for data annotation, algorithm training, model optimization, and achieve data closure has become one of the core competitive advantages that major automakers are vying for.

[0003] CN 112884942 A discloses a data recording and playback system and its playback method. This method includes an autonomous driving domain controller, a host computer, and a sensor data acquisition module. The autonomous driving domain controller includes an adaptive application layer, a data distribution service layer communicating with the adaptive application layer, and a communication management module bidirectionally communicating with the data distribution service layer. The host computer includes a data packet command module, a console communicating with the data packet command module, and a data warehouse theme layer communicating with the data packet command module. The sensor data acquisition module includes LiDAR, ultrasonic radar, millimeter-wave radar, cameras, and a navigation module. However, this solution does not propose a unified framework for abstracting automotive sensor data to adapt to various types and communication methods of sensors. Previous project development has shown that discrete and irregular adaptation of various sensors is a cumbersome, repetitive, and inefficient approach.

[0004] CN 210038149 U utility model relates to a multi-sensor data acquisition system for vehicles. This system includes a lidar, an industrial control computer, multiple cameras, and multiple millimeter-wave radars. The lidar, cameras, and millimeter-wave radars are centrally mounted on a mounting platform located on the roof of the vehicle. The lidar, cameras, and millimeter-wave radars are each connected to the industrial control computer, which is also connected to the vehicle's CAN bus. The lidar is mounted at the center of the mounting platform; multiple cameras are distributed around the lidar on the mounting platform; the multiple millimeter-wave radars, including one forward-facing millimeter-wave radar and multiple angular millimeter-wave radars, are mounted at the perimeter of the mounting platform. This solution primarily focuses on the structural installation design of the sensors, and the sensors and communication methods involved are relatively fixed and inflexible, making it unsuitable for today's era of rapid software and hardware iteration. Summary of the Invention

[0005] This invention provides a method, apparatus, vehicle, equipment, and medium for iterative vehicle driving algorithm. It utilizes a single software program to achieve rapid adaptation of sensor types and communication methods on different vehicle models, and efficiently uses sensor data to drive the application and iterative updates of driving algorithms for intelligent driving vehicles.

[0006] The technical solution of this invention is as follows:

[0007] This invention provides an iterative method for a vehicle driving algorithm, the method comprising:

[0008] Based on sensor profiles, obtain the raw sensor data required by the vehicle.

[0009] Based on sensor configuration files, the acquired raw sensor data is converted into sensor abstract data in a unified format and then published.

[0010] Send the abstract sensor data required by the cloud to the cloud;

[0011] Receive the updated driving algorithm from the cloud based on sensor abstract data, and perform iterative updates to the driving algorithm;

[0012] The sensor profile is generated by adding or removing sensor instances from historical sensor profiles that have been applied to previous models, based on the sensor configuration requirements of the newly developed vehicle model.

[0013] Furthermore, prior to the step of acquiring the raw sensor data required by the vehicle based on the sensor profile, the method further includes:

[0014] Retrieve the sensor configuration file that has been configured with sensor information in the cloud;

[0015] Based on the sensor configuration file, the sensor configuration is loaded on the vehicle side.

[0016] Furthermore, the sensor configuration file includes:

[0017] Sensor instance types and number, data reception and communication methods, data parsing methods, data packaging methods, topic subscription, topic publication, and topic publication cycle for each sensor instance.

[0018] Furthermore, based on the aforementioned sensor configuration file, the steps for loading sensor configuration on the vehicle side include:

[0019] Identify the sensor instances to be loaded based on the sensor configuration file;

[0020] For each sensor instance to be loaded, initialize the communication interface, initialize the topic subscription, and initialize the topic publication.

[0021] Furthermore, based on sensor profiles, the steps for obtaining the raw sensor data required by the vehicle include:

[0022] According to the subscription topic and data reception communication method recorded in the sensor configuration file, the sensor initial data matching the subscription topic is received from the corresponding data interface.

[0023] Furthermore, based on sensor configuration files, the steps for converting the acquired raw sensor data into sensor abstract data in a unified format and publishing it include:

[0024] The initial data of the sensor is parsed according to the data parsing method recorded in the sensor configuration file;

[0025] The parsed initial sensor data is packaged into sensor abstract data in a unified format according to the data packaging method recorded in the sensor configuration file;

[0026] Sensor abstract data is published according to the publication topic and topic publication cycle recorded in the sensor configuration file.

[0027] Furthermore, after the driving algorithm is iteratively updated, the method also includes:

[0028] Obtain the abstract sensor data required by the vehicle;

[0029] Based on the sensor abstract data required by the vehicle, vehicle control commands are generated using iteratively updated driving algorithms.

[0030] Furthermore, the method also includes:

[0031] Receive data feed files generated by the cloud based on sensor abstract data;

[0032] The data backfill file is input into the driving algorithm that has not been iteratively updated, generating vehicle control test instructions for testers to troubleshoot problems.

[0033] Furthermore, the sensor is:

[0034] One or more of the following: temperature sensor, vehicle information, millimeter-wave radar, ultrasonic radar, lidar, surround-view camera, panoramic camera, rear-view camera, front-view camera, and inertial navigation.

[0035] Furthermore, the data interfaces involved in the data reception communication methods recorded in the sensor configuration file are one or more of the following: MIPI, LVDS, RGB, SPI, EHT, CAN, I2C, and UART.

[0036] Furthermore, the vehicle can receive updated driving algorithms from the cloud based on sensor abstract data via its OTA module.

[0037] This invention also provides a vehicle driving algorithm iteration device, the device comprising:

[0038] The sensor data processing center module is used for: acquiring raw sensor data required by the vehicle based on sensor configuration files; converting the acquired raw sensor data into sensor abstract data in a unified format and publishing it based on sensor configuration files; and sending sensor abstract data required by the cloud to the cloud.

[0039] The driving algorithm application module is used to: receive the updated driving algorithm from the cloud based on sensor abstract data, and perform iterative updates to the driving algorithm;

[0040] The sensor profile is generated by adding or removing sensor instances from historical sensor profiles that have been applied to previous models, based on the sensor configuration requirements of the newly developed vehicle model.

[0041] Furthermore, the device includes:

[0042] The sensor data processing center module is also used for:

[0043] Retrieve the sensor configuration file that has been configured with sensor information in the cloud;

[0044] Based on the sensor configuration file, the sensor configuration is loaded on the vehicle side.

[0045] Furthermore, the driving algorithm application module is also used for:

[0046] Obtain the abstract sensor data required by the vehicle;

[0047] Based on the sensor abstract data required by the vehicle, vehicle control commands are generated using iteratively updated driving algorithms.

[0048] This invention also provides a vehicle, including the vehicle driving algorithm iteration device described above.

[0049] This invention also provides a control device, including a processor, a memory, and a program or instructions stored in the memory and executable on the processor. When the program or instructions are executed by the processor, they implement the steps of the vehicle driving algorithm iteration method described above.

[0050] This invention also provides a readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the vehicle driving algorithm iteration method described above.

[0051] The beneficial effects of this invention are as follows:

[0052] By utilizing sensor configuration files, real-world data collected by the vehicle's actual sensors is transmitted to the cloud. The cloud then updates the current version of the driving algorithm installed on the vehicle according to existing algorithm update methods. The locally stored driving algorithm is then synchronously updated based on the cloud-updated algorithm. For a newly developed vehicle model, especially for new and old versions or high / low-spec versions of the same model with sensor hardware modifications, only the sensor configuration file needs to be updated based on the sensor hardware information of the new model. The new model can then utilize the existing software programs of the older model to perform vehicle control. This approach offers high compatibility with sensor hardware, minimal changes to the software program, high software portability, and is well-suited to the demands of today's rapid software and hardware iteration. Furthermore, by using real-world data collected by the vehicle's actual sensors for driving algorithm updates, a closed-loop data mechanism for automotive sensors is achieved, thereby improving the iteration efficiency and quality of the driving algorithm. Attached Figure Description

[0053] The following is a brief explanation of the content represented by each figure in this specification:

[0054] Figure 1 This is a flowchart of the method according to the first embodiment of the present invention;

[0055] Figure 2 This is a detailed flowchart of the method according to the first embodiment of the present invention;

[0056] Figure 3 This is a flowchart of the method according to the second embodiment of the present invention;

[0057] Figure 4 This is a flowchart of step S202 in the second embodiment of the present invention;

[0058] Figure 5 This is a flowchart of the method in the third embodiment of the present invention;

[0059] Figure 6 This is a flowchart of the method in the fourth embodiment of the present invention;

[0060] Figure 7 This is an overall architecture diagram of the automotive sensor data abstraction architecture in the fifth embodiment of the present invention;

[0061] Figure 8This is a data flow diagram of the automotive sensor abstract architecture in the fifth embodiment of the present invention, which uses automotive sensor data to drive the application iteration of driving algorithms. Detailed Implementation

[0062] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0063] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0064] like Figure 1 As shown, in the first embodiment of the present invention, a vehicle driving algorithm iteration method is provided. This method mainly utilizes sensor configuration files to transmit real data collected by the actual sensors of the required vehicle to the cloud. The cloud then updates the current version of the driving algorithm installed on the vehicle according to existing algorithm update methods. The locally stored driving algorithm is then updated synchronously according to the cloud-updated algorithm. For a newly developed vehicle model, especially for new and old versions or high- and low-spec versions of the same model with sensor hardware modifications, only the sensor configuration file needs to be updated based on the sensor hardware information of the newly developed model. The newly developed model can then use the existing software program of the old model to perform vehicle control. This method has high adaptability to sensor hardware, minimal modifications to the software program, high software program portability, and is well-suited to the current needs of rapid software and hardware iteration. Furthermore, because the driving algorithm update utilizes real data collected by the actual vehicle sensors, a closed loop of automotive sensor data is achieved, thus improving the iteration efficiency and quality of the driving algorithm.

[0065] Specifically, in the first embodiment of the present invention, the method specifically includes:

[0066] Step S101: Based on the sensor configuration file, obtain the raw sensor data required by the vehicle.

[0067] Step S102: Based on the sensor configuration file, convert the acquired raw sensor data into sensor abstract data in a unified format and publish it;

[0068] Step S103: Send the sensor abstract data required by the cloud to the cloud.

[0069] Step S104: Receive the updated driving algorithm from the cloud based on sensor abstract data, and perform iterative updates to the driving algorithm.

[0070] In step S101, the raw sensor data refers to the real data collected by the actual vehicle sensors located on the vehicle.

[0071] The sensor profile is generated by adding or removing sensor instances from historical sensor profiles that have been applied to previous models, based on the sensor configuration requirements of the newly developed vehicle model.

[0072] The relationship between historical models and newly developed models can be between old and new models on the same platform, or between high and low versions of the same model on the same platform; or between two models on different platforms.

[0073] The sensor configuration file contains the following records:

[0074] Sensor instance types and number, data reception and communication methods, data parsing methods, data packaging methods, topic subscription, topic publication, and topic publication cycle for each sensor instance.

[0075] Combination Figure 6 In this first embodiment, the supported sensor instance types can be at least: temperature sensor, vehicle information sensor, millimeter-wave radar, ultrasonic radar, lidar, surround-view camera, panoramic camera, rear-view camera, front-view camera, and / or inertial navigation. The specific sensors used are designed based on the actual configuration requirements of the newly developed vehicle model. The data interfaces involved in the data reception communication methods recorded in the sensor configuration file can be one or more of the following: MIPI interface, LVDS interface, RGB interface, SPI, EHT interface, CAN interface, I2C interface, and UART interface. For example, the CAN interface can be used to acquire vehicle information, and the I2C interface can be used to acquire data from millimeter-wave radar and ultrasonic radar. The specific interfaces used are designed based on the actual configuration requirements of the newly developed vehicle model.

[0076] The number of sensor instances is determined by the R&D personnel by adding or removing sensor instances from the historical sensor configuration file based on the sensor types installed on the newly developed vehicle model. At the same time, for the same sensor instance, the data reception and communication methods and data types adapted to the new and old vehicle models may change, and the R&D personnel also need to update the specific information of the edited sensor instances based on the actual situation of the newly developed vehicle model.

[0077] like Figure 2In step S101, specifically step S1011 is: according to the subscription topic and data receiving communication method recorded in the sensor configuration file, receive the initial sensor data that matches the subscription topic from the corresponding data interface.

[0078] like Figure 2 The process in step S102 parses and packages data from different sensor types into cloud-recognizable information. Specifically, step S102 includes:

[0079] Step S1021: Parse the initial sensor data according to the data parsing method recorded in the sensor configuration file;

[0080] Step S1022: Package the parsed initial sensor data into sensor abstract data of a unified format according to the data packaging method recorded in the sensor configuration file;

[0081] Step S1023: Publish sensor abstract data according to the publishing topic and topic publishing cycle recorded in the sensor configuration file.

[0082] In step S103, the sensor abstract data required by the cloud is also determined by the sensor topics subscribed to by the cloud. For the cloud, existing technologies are used to perform data annotation (the specific implementation process utilizes existing technologies), scene extraction (the specific implementation process utilizes existing technologies), data analysis (the specific implementation process utilizes existing technologies), algorithm training (the specific implementation process utilizes existing technologies), and testing and verification (the specific implementation process utilizes existing technologies) on the sensor abstract data to obtain an updated version of the driving algorithm.

[0083] The data packaging process in step S1022 can align the time of data from various sensors, facilitating subsequent data fusion processing by the driving algorithm application module based on time synchronization.

[0084] In step S104, the vehicle receives the updated driving algorithm from the cloud based on sensor abstract data via the OTA module.

[0085] The first embodiment of this invention utilizes the above-described processing flow, updating only the sensor configuration file to adapt to the sensor hardware configuration of the developing vehicle (including, but not limited to, temperature sensors, vehicle information (obtained through an interface, abstractly understood as a sensor), millimeter-wave radar, ultrasonic radar, lidar, surround-view cameras, panoramic cameras, rear-view cameras, front-view cameras, inertial navigation, etc.). It allows reuse of existing vehicle applications, improving sensor development and adaptation efficiency. The sensor configuration file is used to read sensor data from the newly developed vehicle, and the data is parsed and packaged to update the driving algorithm application for the new vehicle. This method supports various communication interfaces to acquire raw data from various sensors, including but not limited to MIPI, LVDS, RGB, SPI, UART, CAN / CAN_FD, I2C, UDP, TCP, SomeIP, MQTT, DDS, etc. The cloud can update the vehicle-side sensor configuration file via OTA, enabling online modification of the sensor communication interface. It features expandable communication interfaces and rapid updates. The cloud can update the vehicle's sensor configuration files via OTA (Over-The-Air) updates, changing the number and type of instantiated sensors on the vehicle, and altering the topics published and subscribed to by these sensor instances, offering flexible and diverse configuration options. Furthermore, this method allows for real-time uploading of real-vehicle sensor data to the cloud, where data annotation, scene extraction, algorithm training, model optimization, and algorithm refinement are performed. Ultimately, the driving algorithm on the real vehicle is updated via OTA, achieving a closed loop for automotive sensor data and improving the iteration efficiency and quality of the driving algorithm.

[0086] like Figure 3 The second embodiment of the present invention also provides a vehicle driving algorithm iteration method. Compared with the first embodiment, the second embodiment mainly adds a sensor loading process. Figure 2 The specific processes of steps S203 to S206 are consistent with the specific processes of S101 to S104 in the previous embodiment. For example... Figure 3 The method specifically includes:

[0087] Step S201: Obtain the sensor configuration file that has been configured with sensor information in the cloud;

[0088] Step S202: Based on the sensor configuration file, load the sensor configuration on the vehicle side;

[0089] Step S203: Based on the sensor configuration file, obtain the raw sensor data required by the vehicle.

[0090] Step S204: Based on the sensor configuration file, convert the acquired raw sensor data into sensor abstract data in a unified format and publish it;

[0091] Step S205: Send the sensor abstract data required by the cloud to the cloud.

[0092] Step S206: Receive the updated driving algorithm from the cloud based on sensor abstract data, and perform iterative updates to the driving algorithm.

[0093] In step S201, the aforementioned sensor configuration file is obtained from the cloud via the vehicle's OTA channel. Combined with... Figure 4 In this second embodiment, step S202 specifically includes:

[0094] Step S2021: Read the sensor configuration file. The configuration file includes the sensor instance type and number, sensor instance data receiving communication method, sensor instance data parsing method, sensor instance data packaging method, sensor instance topic subscription, sensor instance topic publishing, and topic publishing cycle.

[0095] Step S2022: Based on the read sensor configuration file, load the sensor instance and load the corresponding data parsing method and data packaging method.

[0096] Step S2023: Based on the read sensor configuration file, initialize the communication interface of the sensor instance to receive raw data output by the actual vehicle sensors.

[0097] Step S2024: Based on the read sensor configuration file, initialize the subscription topic of the sensor instance and determine the data types and content of the sensors to be subscribed to.

[0098] Step S2025: Based on the read sensor configuration file, initialize the publishing topic of the sensor instance and determine the abstract data type and content to be published.

[0099] Step S2026: Repeat steps S2022 to S2025 until all sensor instances are loaded.

[0100] The sensor configuration file can be parsed using steps S2021 to S2026.

[0101] The second embodiment of the present invention achieves the technical effects of the first embodiment, and on this basis, it also realizes how to load the sensor files configured in the cloud onto the vehicle.

[0102] like Figure 5In the third embodiment of the present invention, based on the second embodiment, a process is added whereby the vehicle-side uses the updated driving algorithm to generate vehicle control commands. The implementation process of steps S301 to S306 is consistent with that of the second embodiment. The method in this third embodiment specifically includes:

[0103] Step S301: Obtain the sensor configuration file that has been configured with sensor information in the cloud;

[0104] Step S302: Based on the sensor configuration file, load the sensor configuration on the vehicle side;

[0105] Step S303: Based on the sensor configuration file, obtain the raw sensor data required by the vehicle.

[0106] Step S304: Based on the sensor configuration file, convert the acquired raw sensor data into sensor abstract data in a unified format and publish it;

[0107] Step S305: Send the sensor abstract data required by the cloud to the cloud.

[0108] Step S306: Receive the updated driving algorithm from the cloud based on sensor abstract data, and perform iterative updates to the driving algorithm;

[0109] Step S307: Obtain the sensor abstract data required by the vehicle.

[0110] Step S308: Based on the sensor abstract data required by the vehicle, vehicle control commands are generated using the iteratively updated driving algorithm. In S307 and S308, the driving algorithm application in the driving domain subscribes to the sensor abstract data published by the sensor data processing center, and then performs multi-sensor data fusion (multi-sensor data fusion is an information processing process that uses computer science technology to mathematically analyze and synthesize information and data from various sensors or multiple sources under predetermined rules to complete corresponding decisions and necessary estimations; the specific implementation utilizes existing technologies), environment reconstruction (the specific implementation utilizes existing technologies), and predictive planning (the specific implementation utilizes existing technologies), ultimately outputting lateral and longitudinal vehicle control commands to the vehicle control domain.

[0111] In the vehicle domain, the vehicle control domain receives lateral and longitudinal vehicle control commands from the driving domain and actually executes vehicle control tasks, including speed control, braking control, lighting control, steering angle control, door and window control, voice control, seat control, etc., thus realizing intelligent driving vehicle control functions.

[0112] like Figure 6In the fourth embodiment of the present invention, a technical solution is provided that uses vehicle-side data to generate a data recovery file in the cloud, and the vehicle-side uses the data recovery file to reproduce driving scenarios for remote troubleshooting. The implementation process of steps S401 to S408 is consistent with that of the third embodiment. The method in this fourth embodiment specifically includes:

[0113] Step S401: Obtain the sensor configuration file that has been configured with sensor information in the cloud;

[0114] Step S402: Based on the sensor configuration file, load the sensor configuration on the vehicle side;

[0115] Step S403: Based on the sensor configuration file, obtain the raw sensor data required by the vehicle.

[0116] Step S404: Based on the sensor configuration file, convert the acquired raw sensor data into sensor abstract data in a unified format and publish it;

[0117] Step S405: Send the sensor abstract data required by the cloud to the cloud.

[0118] Step S406: Receive the updated driving algorithm from the cloud based on sensor abstract data, and perform iterative updates to the driving algorithm;

[0119] Step S407: Obtain the sensor abstract data required by the vehicle.

[0120] Step S408: Based on the sensor abstract data required by the vehicle, generate vehicle control commands using the iteratively updated driving algorithm.

[0121] Step S409: Receive the data feedback file generated by the cloud based on the sensor abstract data;

[0122] Step S410: Input the data backfill file into the driving algorithm that has not been iteratively updated, and generate vehicle control test instructions for testers to troubleshoot problems.

[0123] When necessary, the cloud can re-import the data back to the vehicle to recreate the driving scenario and generate new lateral and longitudinal vehicle control commands. By comparing the old and new lateral and longitudinal vehicle control commands, intelligent driving-related issues can be analyzed and identified. Finally, the tested and verified driving algorithm application can be updated and iterated to reproduce the driving scenario, facilitating the rapid reproduction and analysis of driving problems and the optimization of driving algorithm applications.

[0124] like Figure 7In the fifth embodiment of the present invention, in order to implement the method flow of the aforementioned four embodiments, an automotive sensor data abstraction architecture is applied. The overall solution includes five parts: cloud, real vehicle sensor, driving domain, vehicle control domain, and OTA module.

[0125] The cloud platform primarily enables functions such as data subscription, data classification and organization, data anonymization, data storage, data labeling, scene extraction, data algorithm training, data analysis, model optimization, algorithm optimization, sensor configuration file updates, and data re-feedback file generation.

[0126] Vehicle sensors include, but are not limited to, temperature sensors, vehicle information sensors, millimeter-wave radar, ultrasonic radar, lidar, surround-view cameras, panoramic cameras, rear-view cameras, front-view cameras, inertial navigation, etc.

[0127] The driving domain includes a sensor data processing center and driving algorithm applications. The sensor data processing center primarily handles functions such as sensor configuration file parsing, sensor instance loading, data receiving interface configuration, data subscription, data parsing, data packaging, and publishing of abstract sensor data. The driving algorithm applications subscribe to the abstract sensor data published by the sensor data processing center, then perform data fusion, environment reconstruction, prediction and planning, ultimately outputting lateral and longitudinal vehicle control commands to the vehicle control domain.

[0128] The vehicle control domain receives lateral and longitudinal vehicle control commands from the driving domain and actually executes vehicle control tasks, including speed control, braking control, lighting control, steering angle control, door and window control, sound control, seat control, etc., thus realizing intelligent driving vehicle control functions.

[0129] Figure 8 This is a data flow diagram in the fifth embodiment of the automotive sensor data abstraction architecture, which uses automotive sensor data to drive the iteration of driving algorithms. The process specifically includes:

[0130] The first step is for the vehicle's sensors to send raw sensor data through a predetermined communication method.

[0131] The second step involves the sensor data processing center in the driving domain receiving raw data from multiple sensors from the subscribed topics.

[0132] The third step is to parse, package, and publish the abstract sensor data from multiple sensors.

[0133] The fourth step involves the driving domain's driving algorithm application subscribing to and receiving abstract data from sensors, fusing data from multiple sensors, reconstructing the environment, making predictions and plans, and outputting lateral and longitudinal vehicle control commands. Finally, the vehicle control domain executes the vehicle control tasks according to the commands.

[0134] Fifth, the cloud also subscribes to and receives abstract sensor data. On one hand, it performs data annotation, scene extraction, data analysis, algorithm training, and testing on the abstract sensor data, ultimately updating and iterating the driving application algorithms and models via the OTA module. On the other hand, it de-identifies the data, classifies and organizes the data, generates data re-import files, and completes data storage. When needed, the data re-import files can be re-imported into the sensor data processing center to recreate the driving scenario, compare vehicle control commands, analyze and troubleshoot intelligent driving-related issues, and finally update and iterate the tested and verified driving algorithm application.

[0135] This automotive sensor data abstraction framework can achieve the technical effects of the aforementioned four embodiments. That is:

[0136] 1. This framework is compatible with various types of sensors from different manufacturers, including but not limited to temperature sensors, vehicle information sensors, millimeter-wave radar, ultrasonic radar, lidar, surround-view cameras, panoramic cameras, rear-view cameras, front-view cameras, inertial navigation, and more. The framework includes sensor classes and provides methods for configuring sensor profiles, receiving, parsing, packaging, and publishing sensors. It features rapid adaptation, minimal modifications, and high portability, thus improving development and adaptation efficiency.

[0137] 2. This framework supports various communication interfaces to acquire data from various sensors, including but not limited to MIPI, LVDS, RGB, SPI, UART, CAN / CAN_FD, I2C, UDP, TCP, SOMEIP, MQTT, DDS, etc. The cloud can update the vehicle's sensor configuration files via OTA (Over-The-Air) updates, enabling online modification of sensor communication interfaces. It features scalable communication interfaces and rapid updates.

[0138] 3. The cloud-based system of this framework can update the vehicle's sensor configuration files via OTA, change the number and type of instantiated sensors on the vehicle, and change the topic content published and subscribed to by the sensor instances, offering flexible and diverse configuration options.

[0139] 4. This framework can upload real vehicle sensor data to the cloud in real time, perform data annotation, scene extraction, algorithm training, model optimization, and algorithm optimization in the cloud, and finally update the real vehicle's driving algorithm through the OTA channel, realizing a closed loop of automotive sensor data and improving the iteration efficiency and quality of driving algorithms.

[0140] 5. This framework abstracts and manages various data from different sensors, sending the abstract sensor data to the driving algorithm application according to a predetermined topic protocol. The driving algorithm application then performs data fusion, environmental reconstruction, prediction and planning based on the abstract sensor data, and issues lateral and longitudinal vehicle control commands. This automotive sensor data abstraction framework relatively isolates sensors and driving algorithm applications, facilitating platform portability and reuse.

[0141] 6. This framework aligns the time of data from various sensors, facilitating data fusion processing by subsequent driving algorithm application modules based on time synchronization.

[0142] 7. This framework supports the reloading of sensor data stored in the cloud, enabling the reproduction of driving scenarios and facilitating the rapid reproduction and analysis of driving problems and the optimization of driving algorithm applications.

[0143] The fifth embodiment described above can be summarized as: a vehicle driving algorithm iteration device, the device comprising:

[0144] The sensor data processing center module is used for: acquiring raw sensor data required by the vehicle based on sensor configuration files; converting the acquired raw sensor data into sensor abstract data in a unified format and publishing it based on sensor configuration files; and sending sensor abstract data required by the cloud to the cloud.

[0145] The driving algorithm application module is used to: receive the updated driving algorithm from the cloud based on sensor abstract data, and perform iterative updates to the driving algorithm;

[0146] The sensor profile is generated by adding or removing sensor instances from historical sensor profiles that have been applied to previous models, based on the sensor configuration requirements of the newly developed vehicle model.

[0147] Furthermore, the device includes:

[0148] The sensor data processing center module is also used for:

[0149] Retrieve the sensor configuration file that has been configured with sensor information in the cloud;

[0150] Based on the sensor configuration file, the sensor configuration is loaded on the vehicle side.

[0151] Furthermore, the driving algorithm application module is also used for:

[0152] Obtain the abstract sensor data required by the vehicle;

[0153] Based on the sensor abstract data required by the vehicle, vehicle control commands are generated using iteratively updated driving algorithms.

[0154] This invention also provides a vehicle, including the vehicle driving algorithm iteration device described above.

[0155] This invention also provides a control device, including a processor, a memory, and a program or instructions stored in the memory and executable on the processor. When the program or instructions are executed by the processor, they implement the steps of the vehicle driving algorithm iteration method described above.

[0156] This invention also provides a readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of the vehicle driving algorithm iteration method described above.

[0157] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. An iterative method for a vehicle driving algorithm, characterized in that, The method includes: Based on the sensor configuration file, the raw sensor data required by the vehicle is obtained. The sensor configuration file includes: sensor instance type and number, data receiving and communication method, data parsing method, data packaging method, topic subscription, topic publishing, and topic publishing cycle for each sensor instance. The steps for obtaining the raw sensor data required by the vehicle based on the sensor configuration file specifically include: receiving the initial sensor data matching the subscribed topic from the corresponding data interface according to the subscription topic and data receiving communication method recorded in the sensor configuration file; the abstract sensor data required by the cloud is determined by the sensor topic subscribed to by the cloud. Based on sensor configuration files, the acquired raw sensor data is converted into sensor abstract data in a unified format and then published. Send the abstract sensor data required by the cloud to the cloud; Receive the updated driving algorithm from the cloud based on sensor abstract data, and perform iterative updates to the driving algorithm; The sensor profile is generated by adding or removing sensor instances from historical sensor profiles that have been applied to previous models, based on the sensor configuration requirements of the newly developed vehicle model.

2. The vehicle driving algorithm iteration method according to claim 1, characterized in that, Before the step of acquiring the raw sensor data required by the vehicle based on the sensor profile, the method further includes: Retrieve the sensor configuration file that has been configured with sensor information in the cloud; Based on the sensor configuration file, the sensor configuration is loaded on the vehicle side.

3. The vehicle driving algorithm iteration method according to claim 2, characterized in that, Based on the sensor configuration file, the steps for loading the sensor configuration on the vehicle side include: Identify the sensor instances to be loaded based on the sensor configuration file; For each sensor instance to be loaded, initialize the communication interface, initialize the topic subscription, and initialize the topic publication.

4. The vehicle driving algorithm iteration method according to claim 2, characterized in that, Based on sensor configuration files, the steps for converting the acquired raw sensor data into sensor abstract data in a unified format and publishing it include: The initial data of the sensor is parsed according to the data parsing method recorded in the sensor configuration file; The parsed initial sensor data is packaged into sensor abstract data in a unified format according to the data packaging method recorded in the sensor configuration file; Sensor abstract data is published according to the publication topic and topic publication cycle recorded in the sensor configuration file.

5. The vehicle driving algorithm iteration method according to claim 1, characterized in that, After the driving algorithm is iteratively updated, the method further includes: Obtain the abstract sensor data required by the vehicle; Based on the sensor abstract data required by the vehicle, vehicle control commands are generated using iteratively updated driving algorithms.

6. The vehicle driving algorithm iteration method according to claim 1, 2, or 5, characterized in that, The method further includes: Receive data feed files generated by the cloud based on sensor abstract data; The data backfill file is input into the driving algorithm that has not been iteratively updated, generating vehicle control test instructions for testers to troubleshoot problems.

7. The vehicle driving algorithm iteration method according to claim 6, characterized in that, The sensor is: One or more of the following: temperature sensor, vehicle information, millimeter-wave radar, ultrasonic radar, lidar, surround-view camera, panoramic camera, rear-view camera, front-view camera, and inertial navigation.

8. The vehicle driving algorithm iteration method according to claim 1, characterized in that, The data interfaces involved in the data reception communication methods recorded in the sensor configuration file are one or more of the following: MIPI, LVDS, RGB, SPI, EHT, CAN, I2C, and UART.

9. The vehicle driving algorithm iteration method according to claim 1, characterized in that, The vehicle receives updated driving algorithms from the cloud based on sensor abstraction data via its OTA module.

10. A vehicle driving algorithm iteration device, characterized in that, The device includes: The sensor data processing center module is used for: acquiring raw sensor data required by the vehicle based on sensor configuration files; converting the acquired raw sensor data into sensor abstract data in a unified format and publishing it based on sensor configuration files; and sending sensor abstract data required by the cloud to the cloud. The sensor configuration file includes: sensor instance type and number, data receiving and communication methods, data parsing methods, data packaging methods, topic subscription, topic publishing, and topic publishing cycle for each sensor instance. The steps for obtaining the raw sensor data required by the vehicle based on the sensor configuration file specifically include: receiving the initial sensor data matching the subscribed topic from the corresponding data interface according to the subscription topic and data receiving communication method recorded in the sensor configuration file; the abstract sensor data required by the cloud is determined by the sensor topic subscribed to by the cloud. The driving algorithm application module is used to: receive the updated driving algorithm from the cloud based on sensor abstract data, and perform iterative updates to the driving algorithm; The sensor profile is generated by adding or removing sensor instances from historical sensor profiles that have been applied to previous models, based on the sensor configuration requirements of the newly developed vehicle model.

11. The vehicle driving algorithm iteration device according to claim 10, characterized in that, The device includes: The sensor data processing center module is also used for: Retrieve the sensor configuration file that has been configured with sensor information in the cloud; Based on the sensor configuration file, the sensor configuration is loaded on the vehicle side.

12. The vehicle driving algorithm iteration device according to claim 10 or 11, characterized in that, The driving algorithm application module is also used for: Obtain the abstract sensor data required by the vehicle; Based on the sensor abstract data required by the vehicle, vehicle control commands are generated using iteratively updated driving algorithms.

13. A vehicle, characterized in that, Includes the vehicle driving algorithm iteration device according to any one of claims 9 to 12.

14. A control device, characterized in that, It includes a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions, when executed by the processor, implement the steps of the vehicle driving algorithm iterative method as described in any one of claims 1 to 8.

15. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the vehicle driving algorithm iterative method as described in any one of claims 1 to 8.

Images