Data radar cooperative automobile automatic driving control method and system based on dds protocol

Abstract

This invention discloses a data radar-coordinated vehicle autonomous driving control method and system based on the DDS protocol. The method includes: S1: completing the registration and configuration of the data radar model; S2: the data radar processes raw data from multiple sources, and obtains target state information through format unification, time synchronization, and Kalman filtering; S3: the target information is compressed and encrypted and published through DDS, with roadside terminals subscribing and transmitting it to the regional cloud control center; S4: the cloud control center combines vehicle condition information and uses reinforcement learning to generate driving guidance instructions; S5: the instructions are parsed back by DDS and converted into CAN bus control signals to control the vehicle; S6-S8: respectively realizing dynamic updating of DCPS parameters, topic information caching, and link detection switching; this invention achieves multi-level seamless integration through the DDS architecture and DCPS mechanism, solving the interaction barriers of heterogeneous systems; the optimized algorithm improves perception accuracy and decision-making scientificity; dynamic updates, caching, and redundancy switching enhance the robustness and adaptability of the system.

Description

Technical Field

[0001] This invention relates to the field of data radar modeling technology, specifically to a data radar-coordinated autonomous driving control method and system for automobiles based on the DDS protocol. Background Technology

[0002] With the rapid development of intelligent connected vehicle technology, the requirements for real-time data interaction, multi-device collaboration, and dynamic decision-making in autonomous driving systems are increasing. Among these, the data transmission efficiency between the vehicle perception layer and the cloud control center, the fusion accuracy of multi-source sensor information, and the response speed of decision commands under complex road conditions directly affect the safety and reliability of autonomous driving. Current technologies still have certain shortcomings in these areas.

[0003] Firstly, the vehicle perception layer includes various heterogeneous devices such as cameras, V2X devices, and millimeter-wave radar. The data formats and communication protocols of these devices are not uniform, making it difficult to efficiently integrate the perception data and providing complete and consistent environmental and vehicle status information for subsequent decision-making.

[0004] Secondly, traditional data fusion algorithms do not fully consider the reliability differences in the data transmission process, and decision-making algorithms do not deeply integrate with the real-time status of vehicle hardware. This can easily lead to problems such as insufficient perception accuracy or mismatch between decision commands and the actual operating conditions of the vehicle, affecting the safety of autonomous driving.

[0005] Third, during the operation of an autonomous driving system, the parameters of on-board equipment may be dynamically adjusted, and the communication link may be easily interrupted due to road conditions or signal interference. However, the existing system lacks a flexible parameter adaptation mechanism and redundant link protection scheme, which makes the system prone to operation interruption in dynamically changing environments, resulting in insufficient robustness and environmental adaptability.

[0006] Therefore, it is essential to design a data radar-coordinated autonomous driving control method and system based on the DDS protocol. Summary of the Invention

[0007] The purpose of this invention is to provide a data radar-coordinated autonomous driving control method and system based on the DDS protocol. This addresses the issues raised in the background section, which describes a vehicle perception layer comprising various heterogeneous devices such as cameras, V2X equipment, and millimeter-wave radar. The inconsistent data formats and communication protocols of these devices make efficient integration of perception data difficult, hindering the provision of complete and consistent environmental and vehicle status information for subsequent decision-making. Furthermore, traditional data fusion algorithms do not adequately consider reliability differences during data transmission, and decision-making algorithms do not deeply integrate with the real-time vehicle hardware status, leading to insufficient perception accuracy or mismatch between decision commands and actual vehicle operating conditions, thus affecting autonomous driving safety. Additionally, during autonomous driving system operation, vehicle equipment parameters may dynamically adjust, and communication links are easily interrupted due to road conditions or signal interference. Existing systems lack flexible parameter adaptation mechanisms and redundant link protection schemes, resulting in system interruptions in dynamically changing environments and insufficient robustness and environmental adaptability.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] Firstly, a data radar-assisted autonomous driving control method for vehicles based on the DDS protocol is provided, including the following steps:

[0010] S1: The management platform maintains the configuration information for intelligent connected vehicles. The management platform synchronizes the configuration information to the associated database of the DDS adapter. The DDS adapter extracts the parameter information required by the DCPS object from the database, calls its own API interface to create the DCPS object, and stores the DCPS object information according to domain, QoS parameters, and topic name to achieve decoupling. Data writers and data readers are generated by combining information to complete the registration and configuration of the data radar model in the intelligent connected vehicle system.

[0011] S2: The data radar receives raw data collected by the camera, V2X device and vehicle radar of the intelligent connected vehicle. After the raw data is formatted and time synchronized, the Kalman filter algorithm is used to process it to obtain the fused target state information, which includes the target position, speed and motion trajectory.

[0012] S3: The data radar compresses and encrypts the fused target status information and publishes it to the vehicle-mounted DDS domain through the DCPS object of the DDS adapter with a preset topic as the identifier; the roadside signal receiving terminal subscribes to the topic to obtain the target status information and transmits it to the regional cloud control center through the backhaul network relay.

[0013] S4: The regional cloud control center receives the target status information transmitted by the roadside signal receiving terminal, combines it with the vehicle status information reported by the intelligent connected vehicle through the DDS protocol, and uses reinforcement learning algorithm to generate driving guidance instructions.

[0014] S5: Driving guidance instructions are published to the DDS domain via the roadside signal receiving terminal. The data radar subscribes to the instruction topic through its parsing unit and then parses the instructions, converting them into control signals that conform to the CAN bus standard. The vehicle control unit receives the control signals and controls the vehicle's power system, braking system, and steering system.

[0015] S6: Monitor parameter changes of DCPS objects in real time and update the corresponding level of DCPS objects according to the change type.

[0016] As a further technical solution of the present invention, the configuration information in S1 includes:

[0017] Interface version structure: The format is "major version number, minor version number";

[0018] Region code: uint8 type, value 0-255;

[0019] Country code structure: stores the country name and ISO3166-1alpha-2 encoding;

[0020] Vehicle dynamic information structure: stores vehicle speed, acceleration, and driving direction angle;

[0021] Sensor environmental information structure: stores camera frame rate, radar detection range, and V2X signal strength;

[0022] Control instruction structure: stores instruction type, parameters, and execution time limit;

[0023] DCPS object decoupling is achieved through categorized storage of "domain-QoS parameters-topic name". The value range of the domain is 0-231, and the QoS parameters include participant QoS, publisher QoS, subscriber QoS, topic QoS, data writer QoS, and data reader QoS.

[0024] As a further technical solution of the present invention, in S2, the Kalman filtering algorithm is as follows:

[0025] Prediction phase:

[0026]

[0027]

[0028] in, The target state prediction at time k is based on the observation at time k-1, and includes position and velocity components; The state transition matrix describes the target's trajectory under uniform linear motion. , The sampling time interval; This is the optimal estimate of the target state at time k-1; To control the input matrix; This is the control input at time k-1; Let be the prediction error covariance matrix at time k based on time k-1; Let be the optimal error covariance matrix at time k-1; State transition matrix The transpose of the matrix; This is the process noise covariance matrix, reflecting the uncertainty of the target motion;

[0029] Update phase:

[0030]

[0031]

[0032]

[0033] in, Let K be the Kalman gain matrix at time k; The observation matrix maps state variables to observations. Observation matrix The transpose of the matrix; To observe the noise covariance, which reflects the sensor's accuracy; This is the optimal estimate of the target state at time k; For the introduction of DDS data reliability parameters; Let k be the sensor observation value at time k; Let be the optimal error covariance matrix at time k;

[0034] The reliability level is determined by two core indicators during DDS data transmission: packet loss rate and latency fluctuation. A packet loss rate ≤ 0.1% and latency fluctuation ≤ 5ms is considered "reliable". =0.8-0.9; When the packet loss rate is ≥5% and the latency fluctuation is ≥50ms, it is judged as "best effort" level, corresponding to =0.1-0.3; when the packet loss rate and latency fluctuation are between the above two, it is judged as "semi-reliable" level, corresponding to... =0.4-0.7; the aforementioned The output is the target state information after fusion.

[0035] As a further technical solution of the present invention, in S3, the roadside signal receiving terminal uses multi-antenna MIMO technology and OFDM signal enhancement algorithm to receive data published by the vehicle DDS domain; and uploads the data to the regional cloud control center according to the preset data priority through traffic shaping technology, wherein the priority of emergency braking information is 1, the priority of road condition warning information is 2, the priority of normal status information is 3, and the data transmission delay is: priority 1 ≤ priority 2 ≤ priority 3.

[0036] During DDS transmission, a QoS parameter optimization strategy is adopted, introducing a domain priority P, which is a normalized parameter with a value range of 0-1. Specifically, P = 0.9-1.0 for the vehicle safety domain, P = 0.6-0.8 for the vehicle comfort domain, and P = 0.3-0.5 for the vehicle entertainment domain. The vehicle safety domain includes functions related to collision warning and emergency braking; the vehicle comfort domain includes functions related to air conditioning adjustment and seat control; and the vehicle entertainment domain includes functions related to audio / video playback and navigation entertainment. The transmission delay control formula is as follows:

[0037]

[0038] in, The total transmission latency of data from the data radar to the regional cloud control center; This refers to the data transmission latency on the physical link, which includes the connection between the vehicle and the roadside, and between the roadside and the cloud control center. for The corresponding target state information dataset size, The compression ratio of the target state information; For the data packet processing delay between the data radar and the roadside terminal; This is the delay adjustment factor, which is dynamically determined based on the network congestion level.

[0039] As a further technical solution of the present invention, in S4, the regional cloud control center uses HDFS to store target status information, MongoDB NoSQL database to store vehicle condition information, and Spark framework to implement parallel processing.

[0040] Introducing vehicle health H using reinforcement learning algorithms:

[0041]

[0042] in, For the braking system weight, For battery weight, For tire weight, and ; This is a sub-item for braking system health, calculated based on brake pedal wear rate; This is a sub-item for battery health, calculated based on the remaining battery power. This is a sub-item for tire health, calculated based on tire pressure deviation.

[0043] The reward function formula is:

[0044]

[0045] Where γ is the safety weight, For based on Calculate the collision risk coefficient based on the target location and trajectory. As an energy efficiency weight, For based on The target speed is used to adjust the vehicle's energy consumption strategy.

[0046] As a further technical solution of the present invention, in S5, the data radar parsing unit is deployed on the vehicle MCU;

[0047] The vehicle control unit includes: a power control submodule that controls the engine or motor speed, a brake control submodule that controls the brake master cylinder pressure, and a steering control submodule that controls the EPS steering motor. The control signals strictly comply with the CAN bus communication standard.

[0048] As a further technical solution of the present invention, in S6, the update logic for topic information change is as follows: first, a new topic carrying updated information is created, then the original data reader or writer is re-associated with the new topic, and finally the original topic is deleted;

[0049] The update logic for participant changes is as follows: first, create a new participant with updated parameters, then migrate the original publisher or subscriber and topic to the new participant, and finally cancel the original participant;

[0050] QoS changes for data readers or writers only update the corresponding object, while domain changes require updating all associated participants, topics, publishers or subscribers, and data readers or writers.

[0051] As a further technical solution of the present invention, it also includes S7: using a distributed cache unit to cache data readers, data writers and associated system numbers and service numbers, supporting the management platform to update DCPS objects in real time, and at the same time realizing message transmission according to a specified path;

[0052] The distributed caching unit uses a Redis cluster with a master-slave backup architecture, and is divided into independent cache partitions by domain; the cache update strategy process is as follows:

[0053] ① Create a cached copy when an update request is received;

[0054] ② Perform the update operation on the copy;

[0055] ③ Switch cache access to the replica;

[0056] ④ Delete old cached data.

[0057] As a further technical solution of the present invention, it also includes S8: continuously sending link detection messages to the data radar, roadside signal receiving and transmitting terminal and regional cloud control center through the DDS adapter to verify the communication link status; if no detection response is received, the link is determined to be interrupted, and the automatic switching from the main 5G link to the backup WIFI6 link is triggered to ensure the continuity of data transmission.

[0058] Secondly, it provides a data radar-assisted autonomous driving control system based on the DDS protocol, including:

[0059] Management platform module: Deployed on a cloud server, it includes a configuration management sub-module that maintains the configuration information described in the first aspect and supports visual editing on the web interface, and an API call sub-module that implements data radar model registration and configuration;

[0060] Perception layer module: includes hardware unit and data fusion unit. The hardware unit includes one wide-angle camera on the windshield + two narrow-angle cameras in the rearview mirrors, an on-board OBU supporting the C-V2XPC5 protocol and a roadside RSU supporting 5G+DSRC dual mode, one forward-facing lidar + four four-corner millimeter-wave radars. The data fusion unit is deployed on the data radar and executes the Kalman filtering algorithm described in the first aspect.

[0061] DDS Transmission Module: Includes a DDS adapter and a DCPS object unit. The DDS adapter integrates an ARM Cortex-A53 processor and has a built-in database query submodule for interfacing with MySQL and a DDS-REST protocol conversion submodule. The DCPS object unit is used to divide storage partitions by domain, QoS, and topic.

[0062] Regional cloud control center module: Deployed on an edge cloud server cluster, including a distributed storage unit and an intelligent decision-making unit, running the improved reinforcement learning algorithm described in the first aspect;

[0063] On-board execution module: includes a data radar analysis unit and a vehicle control unit. The data radar analysis unit is deployed on the on-board MCU. The vehicle control unit includes a power control submodule, a braking control submodule and a steering control submodule.

[0064] Theme Management Module: Deployed on the roadside server, it includes a parameter monitoring submodule and a hierarchical update submodule. The parameter monitoring submodule is used to monitor DCPS object parameters, and the hierarchical update submodule is used to execute the update logic described in the first aspect.

[0065] Cache module: Adopts the Redis cluster architecture described in the first aspect, including a partition management submodule and an update strategy submodule. The partition management submodule is used to divide partitions by domain, and the update strategy submodule is used to execute cache update strategies.

[0066] Communication verification module: includes a link detection submodule and a link switching submodule, wherein the link detection submodule is used to send the detection message described in the first aspect, and the link switching submodule is used to execute the link switching logic described in the first aspect;

[0067] Communication relationships between modules: The management platform communicates with the DDS transmission module via the REST protocol; the perception layer communicates with the DDS transmission module via the DDS protocol; the DDS transmission module communicates with the regional cloud control center via a 5G or fiber optic link; the regional cloud control center communicates with the vehicle execution module via the DDS protocol; and the topic management, caching, and communication verification modules communicate with the DDS transmission module via an internal bus, forming a closed loop throughout the entire process.

[0068] Compared with the prior art, the beneficial effects of the data radar cooperative vehicle autonomous driving control method and system based on the DDS protocol of the present invention are:

[0069] By constructing a data radar collaborative architecture based on the DDS protocol and combining it with the standardized registration and configuration mechanism of DCPS objects, seamless integration of the perception layer, transmission layer and decision layer of intelligent connected vehicles is achieved; the unified processing of multi-source sensor information by data radar and the protocol conversion capability of the DDS adapter effectively solve the data interaction barriers between heterogeneous systems and improve the overall data interaction efficiency and collaborative capability of the system.

[0070] Adopting DDS data reliability parameters The Kalman filter algorithm is used to fuse multi-source data, and combined with reinforcement learning algorithms to generate driving commands, so as to synergistically optimize the accuracy of perception layer data and the scientific nature of decision layer commands; by adjusting the reliability parameters of DDS data... By introducing an algorithm model with vehicle health H, the entire chain of intelligence from environmental perception to vehicle control is realized, which significantly improves the safety and decision rationality of the autonomous driving system.

[0071] The introduction of dynamic theme configuration updates, distributed caching, and communication link redundancy switching mechanisms ensures the system's adaptive adjustment capability when parameters change or links malfunction. Real-time monitoring and rapid response mechanisms reduce the risk of system interruption, enhance the robustness and environmental adaptability of the autonomous driving system in complex environments, and ensure the continuity and stability of vehicle control. Attached Figure Description

[0072] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0073] Figure 2 Application architecture diagram for enabling autonomous driving using data radar models;

[0074] Figure 3 Adaptation diagram for communication services of data radar model;

[0075] Figure 4 Registration diagram for data radar model information;

[0076] Figure 5 This is a schematic diagram of the DCPS relationship. Detailed Implementation

[0077] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0078] Please see the appendix Figure 1 Embodiment 1 of the present invention provides a data radar-coordinated vehicle autonomous driving control method based on the DDS protocol, comprising the following steps:

[0079] S1: The management platform maintains the configuration information for intelligent connected vehicles, and synchronizes the configuration information to the associated database of the DDS adapter. The DDS adapter extracts the parameter information required for the DCPS object from the database, calls its own API interface to create the DCPS object, and stores the DCPS object information by domain, QoS parameters, and topic name to achieve decoupling. Data writers and data readers are generated through information combination, completing the registration and configuration of the data radar model in the intelligent connected vehicle system. (See Appendix) Figure 4 ;

[0080] Configuration information includes:

[0081] Interface version structure: The format is "major version number, minor version number";

[0082] Region code: uint8 type, value 0-255;

[0083] Country code structure: stores the country name and ISO3166-1alpha-2 encoding;

[0084] Vehicle dynamic information structure: stores vehicle speed, acceleration, and driving direction angle;

[0085] Sensor environmental information structure: stores camera frame rate, radar detection range, and V2X signal strength;

[0086] Control instruction structure: stores instruction type, parameters, and execution time limit;

[0087] DCPS object decoupling is achieved through "domain-QoS parameter-topic name" categorized storage. The value range of the domain is 0-231, and the QoS parameters include participant QoS, publisher QoS, subscriber QoS, topic QoS, data writer QoS, and data reader QoS.

[0088] S2: The data radar receives raw data collected by the cameras, V2X devices, and vehicle radar of the intelligent connected vehicle. After format unification and time synchronization processing of the raw data, a Kalman filter algorithm is used to process the data to obtain fused target state information, including target position, velocity, and trajectory. (See appendix for details.) Figure 2 ;

[0089] The Kalman filter algorithm is as follows:

[0090] Prediction phase:

[0091]

[0092]

[0093] in, The target state prediction at time k is based on the observation at time k-1, and includes position and velocity components; The state transition matrix describes the target's trajectory under uniform linear motion. , The sampling time interval; This is the optimal estimate of the target state at time k-1; To control the input matrix; This is the control input at time k-1; Let be the prediction error covariance matrix at time k based on time k-1; Let be the optimal error covariance matrix at time k-1; State transition matrix The transpose of the matrix; This is the process noise covariance matrix, reflecting the uncertainty of the target motion;

[0094] Update phase:

[0095]

[0096]

[0097]

[0098] in, Let K be the Kalman gain matrix at time k; The observation matrix maps state variables to observations. Observation matrix The transpose of the matrix; To observe the noise covariance, which reflects the sensor's accuracy; This is the optimal estimate of the target state at time k; For the introduction of DDS data reliability parameters; Let k be the sensor observation value at time k; Let be the optimal error covariance matrix at time k;

[0099] The reliability level is determined by two core indicators during DDS data transmission: packet loss rate and latency fluctuation. A packet loss rate ≤ 0.1% and latency fluctuation ≤ 5ms is considered "reliable". =0.8-0.9; When the packet loss rate is ≥5% and the latency fluctuation is ≥50ms, it is judged as "best effort" level, corresponding to =0.1-0.3; when the packet loss rate and latency fluctuation are between the above two, it is judged as "semi-reliable" level, corresponding to... =0.4-0.7; As the output of the fused target state information;

[0100] S3: The data radar compresses and encrypts the fused target status information and publishes it to the vehicle-mounted DDS domain through the DCPS object of the DDS adapter with a preset topic as the identifier; the roadside signal receiving terminal subscribes to the topic to obtain the target status information and transmits it to the regional cloud control center through the backhaul network relay.

[0101] The roadside signal receiving terminal uses multi-antenna MIMO technology and OFDM signal enhancement algorithms to receive data published by the vehicle-mounted DDS domain; it then uploads the data to the regional cloud control center according to preset data priorities using traffic shaping technology. Emergency braking information has a priority of 1, road condition warning information has a priority of 2, and routine status information has a priority of 3. Data transmission latency is: priority 1 ≤ priority 2 ≤ priority 3. (See appendix for details.) Figure 3 ;

[0102] During DDS transmission, a QoS parameter optimization strategy is adopted, introducing a domain priority P, which is a normalized parameter with a value range of 0-1. Specifically, P = 0.9-1.0 for the vehicle safety domain, 0.6-0.8 for the vehicle comfort domain, and 0.3-0.5 for the vehicle entertainment domain. The vehicle safety domain includes functions related to collision warning and emergency braking; the vehicle comfort domain includes functions related to air conditioning adjustment and seat control; and the vehicle entertainment domain includes functions related to audio / video playback and navigation. The transmission delay control formula is as follows:

[0103]

[0104] in, The total transmission latency of data from the data radar to the regional cloud control center; This refers to the data transmission latency on the physical link, which includes the connection between the vehicle and the roadside, and between the roadside and the cloud control center. for The corresponding target state information dataset size, The compression ratio of the target state information; For the data packet processing delay between the data radar and the roadside terminal; This is a delay adjustment factor, dynamically determined based on network congestion levels;

[0105] S4: The regional cloud control center receives the target status information transmitted by the roadside signal receiving terminal, combines it with the vehicle status information reported by the intelligent connected vehicle through the DDS protocol, and uses reinforcement learning algorithm to generate driving guidance instructions.

[0106] The regional cloud control center uses HDFS to store target status information, MongoDB NoSQL database to store vehicle condition information, and Spark framework to implement parallel processing.

[0107] Introducing vehicle health H using reinforcement learning algorithms:

[0108]

[0109] in, For the braking system weight, For battery weight, For tire weight, and ; This is a sub-item for braking system health, calculated based on brake pedal wear rate; This is a sub-item for battery health, calculated based on the remaining battery power. This is a sub-item for tire health, calculated based on tire pressure deviation.

[0110] The reward function formula is:

[0111]

[0112] Where γ is the safety weight, For based on Calculate the collision risk coefficient based on the target location and trajectory. As an energy efficiency weight, For based on Adjust the vehicle's energy consumption strategy according to the target speed;

[0113] S5: Driving guidance instructions are published to the DDS domain via the roadside signal receiving terminal. The data radar subscribes to the instruction topic through its parsing unit and then parses the instructions, converting them into control signals that conform to the CAN bus standard. The vehicle control unit receives the control signals and controls the vehicle's power system, braking system, and steering system.

[0114] The data radar analysis unit is deployed in the vehicle's MCU;

[0115] The vehicle control unit includes: a power control submodule that controls the speed of the engine or motor, a brake control submodule that controls the pressure of the master cylinder, and a steering control submodule that controls the EPS steering motor. The control signals strictly comply with the CAN bus communication standard.

[0116] S6: Monitors parameter changes of DCPS objects in real time and updates the corresponding level of DCPS objects according to the change type. See Appendix. Figure 5 ;

[0117] The update logic for topic information changes is as follows: first, create a new topic with updated information; then, re-associate the original data readers or writers with the new topic; finally, delete the original topic.

[0118] The update logic for participant changes is as follows: first, create a new participant with updated parameters, then migrate the original publisher or subscriber and topic to the new participant, and finally cancel the original participant;

[0119] QoS changes for data readers or writers only update the corresponding object, while domain changes require updating all associated participants, topics, publishers or subscribers, and data readers or writers.

[0120] It also includes S7: It uses a distributed caching unit to cache data readers, data writers and associated system numbers and service numbers, supports the management platform to update DCPS objects in real time, and enables messages to be transmitted along a specified path;

[0121] The distributed caching unit uses a Redis cluster with a master-slave backup architecture, and is divided into independent cache partitions by domain; the cache update strategy process is as follows:

[0122] ① Create a cached copy when an update request is received;

[0123] ② Perform the update operation on the copy;

[0124] ③ Switch cache access to the replica;

[0125] ④ Delete old cached data;

[0126] It also includes S8: continuously sending link detection messages to the data radar, roadside signal receiving and transmitting terminal and regional cloud control center through the DDS adapter to verify the communication link status; if no detection response is received, the link is determined to be interrupted, triggering automatic switching from the main 5G link to the backup WIFI6 link to ensure data transmission continuity.

[0127] Embodiment 2 provided by the present invention: A data radar-coordinated autonomous driving control system for automobiles based on the DDS protocol, including a management platform module: deployed on a cloud server, containing a configuration management sub-module that maintains the configuration information of Embodiment 1 and supports visual editing on the web, and an API call sub-module that realizes the registration and configuration of the data radar model;

[0128] Perception layer module: includes hardware unit and data fusion unit. The hardware unit includes 1 wide-angle camera on the windshield + 2 narrow-angle cameras in the rearview mirror, an on-board OBU supporting the C-V2XPC5 protocol and a roadside RSU supporting 5G+DSRC dual mode, 1 forward-facing lidar + 4 four-corner millimeter-wave radars. The data fusion unit is deployed on the data radar and executes the Kalman filtering algorithm in Example 1.

[0129] DDS Transmission Module: Includes DDS adapter and DCPS object unit. The DDS adapter integrates an ARM Cortex-A53 processor and has a built-in database query submodule for MySQL and a DDS-REST protocol conversion submodule. The DCPS object unit is used to divide storage partitions by domain, QoS, and topic.

[0130] Regional cloud control center module: Deployed on an edge cloud server cluster, including a distributed storage unit and an intelligent decision-making unit, running the improved reinforcement learning algorithm of Implementation Example 1;

[0131] On-board execution module: includes a data radar analysis unit and a vehicle control unit. The data radar analysis unit is deployed on the on-board MCU. The vehicle control unit includes a power control submodule, a braking control submodule and a steering control submodule.

[0132] Topic Management Module: Deployed on the roadside server, it includes a parameter monitoring submodule and a hierarchical update submodule. The parameter monitoring submodule is used to monitor DCPS object parameters, and the hierarchical update submodule is used to execute the update logic of Example 1.

[0133] Cache module: Adopts the Redis cluster architecture of Example 1, including a partition management submodule and an update strategy submodule. The partition management submodule is used to divide partitions by domain, and the update strategy submodule is used to execute cache update strategies.

[0134] Communication verification module: includes a link detection submodule and a link switching submodule. The link detection submodule is used to send the detection message of Example 1, and the link switching submodule is used to execute the link switching logic of Example 1.

[0135] Communication relationships between modules: The management platform communicates with the DDS transmission module via the REST protocol; the perception layer communicates with the DDS transmission module via the DDS protocol; the DDS transmission module communicates with the regional cloud control center via a 5G or fiber optic link; the regional cloud control center communicates with the vehicle execution module via the DDS protocol; and the topic management, caching, and communication verification modules communicate with the DDS transmission module via an internal bus, forming a closed loop throughout the entire process.

[0136] In summary, this invention achieves seamless integration of the perception, transmission, and decision layers of intelligent connected vehicles by constructing a data radar collaborative architecture based on the DDS protocol and combining it with the standardized registration and configuration mechanism of DCPS objects. The unified processing of multi-source sensor information by the data radar and the protocol conversion capability of the DDS adapter effectively solve the data interaction barriers between heterogeneous systems and improve the overall data interaction efficiency and collaborative capability of the system.

[0137] Adopting DDS data reliability parameters The Kalman filter algorithm is used to fuse multi-source data, and combined with reinforcement learning algorithms to generate driving commands, so as to synergistically optimize the accuracy of perception layer data and the scientific nature of decision layer commands; by adjusting the reliability parameters of DDS data... By introducing an algorithm model with vehicle health H, the entire chain of intelligence from environmental perception to vehicle control is realized, which significantly improves the safety and decision rationality of the autonomous driving system.

[0138] The introduction of dynamic theme configuration updates, distributed caching, and communication link redundancy switching mechanisms ensures the system's adaptive adjustment capability when parameters change or links malfunction. Real-time monitoring and rapid response mechanisms reduce the risk of system interruption, enhance the robustness and environmental adaptability of the autonomous driving system in complex environments, and ensure the continuity and stability of vehicle control.

[0139] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A data radar-coordinated autonomous driving control method for vehicles based on the DDS protocol, characterized in that: Includes the following steps: S1: The management platform maintains the configuration information adapted to intelligent connected vehicles, and the management platform synchronizes the configuration information to the associated database of the DDS adapter; The DDS adapter extracts the parameter information required for the DCPS object from the database, calls its own API interface to create the DCPS object, and stores the DCPS object information according to domain, QoS parameters and topic name to achieve decoupling; it generates data writers and data readers by combining information, and completes the registration and configuration of the data radar model in the intelligent connected vehicle system. S2: The data radar receives raw data collected by the camera, V2X device and vehicle radar of the intelligent connected vehicle. After the raw data is formatted and time synchronized, the Kalman filter algorithm is used to process it to obtain the fused target state information, which includes the target position, speed and motion trajectory. S3: The data radar compresses and encrypts the fused target status information and publishes it to the vehicle-mounted DDS domain through the DCPS object of the DDS adapter with a preset topic as the identifier; the roadside signal receiving terminal subscribes to the topic to obtain the target status information and transmits it to the regional cloud control center through the backhaul network relay. S4: The regional cloud control center receives the target status information transmitted by the roadside signal receiving terminal, combines it with the vehicle status information reported by the intelligent connected vehicle through the DDS protocol, and uses reinforcement learning algorithm to generate driving guidance instructions. S5: Driving guidance instructions are published to the DDS domain via the roadside signal receiving terminal. The data radar subscribes to the instruction topic through its parsing unit and then parses the instructions, converting them into control signals that conform to the CAN bus standard. The vehicle control unit receives the control signals and controls the vehicle's power system, braking system, and steering system. S6: Monitor parameter changes of DCPS objects in real time and update the corresponding level of DCPS objects according to the change type.

2. The data radar-coordinated autonomous driving control method for vehicles based on the DDS protocol according to claim 1, characterized in that: The configuration information in S1 includes: Interface version structure: The format is "major version number, minor version number"; Region code: uint8 type, value 0-255; Country code structure: stores the country name and ISO3166-1alpha-2 encoding; Vehicle dynamic information structure: stores vehicle speed, acceleration, and driving direction angle; Sensor environmental information structure: stores camera frame rate, radar detection range, and V2X signal strength; Control instruction structure: stores instruction type, parameters, and execution time limit; DCPS object decoupling is achieved through "domain-QoS parameter-topic name" categorized storage. The domain value range is 0-231, and the QoS parameters include participant QoS, publisher QoS, subscriber QoS, topic QoS, data writer QoS, and data reader QoS.

3. The data radar-coordinated autonomous driving control method for vehicles based on the DDS protocol according to claim 1, characterized in that: In S2, the Kalman filter algorithm is as follows: Prediction phase: in, The target state prediction at time k is based on the observation at time k-1, and includes position and velocity components; The state transition matrix describes the target's trajectory under uniform linear motion. , The sampling time interval; This is the optimal estimate of the target state at time k-1; To control the input matrix; This is the control input at time k-1; Let be the prediction error covariance matrix at time k based on time k-1; Let be the optimal error covariance matrix at time k-1; State transition matrix The transpose of the matrix; This is the process noise covariance matrix, reflecting the uncertainty of the target motion; Update phase: in, Let K be the Kalman gain matrix at time k; The observation matrix maps state variables to observations. Observation matrix The transpose of the matrix; To observe the noise covariance, which reflects the sensor's accuracy; This is the optimal estimate of the target state at time k; For the introduction of DDS data reliability parameters; Let k be the sensor observation value at time k; Let be the optimal error covariance matrix at time k; The reliability level is determined by two core indicators during DDS data transmission: packet loss rate and latency fluctuation. A packet loss rate ≤ 0.1% and latency fluctuation ≤ 5ms is considered "reliable". =0.8-0.9; When the packet loss rate is ≥5% and the latency fluctuation is ≥50ms, it is judged as "best effort" level, corresponding to =0.1-0.3; when the packet loss rate and latency fluctuation are between the above two, it is judged as "semi-reliable" level, corresponding to... =0.4-0.7; the aforementioned The output is the target state information after fusion.

4. The data radar-coordinated autonomous driving control method for vehicles based on the DDS protocol according to claim 1, characterized in that: In S3, the roadside signal receiving terminal uses multi-antenna MIMO technology and OFDM signal enhancement algorithm to receive data published by the vehicle DDS domain; and uploads it to the regional cloud control center according to preset data priority through traffic shaping technology, wherein emergency braking information has priority 1, road condition warning information has priority 2, and normal status information has priority 3. The data transmission delay is: priority 1 ≤ priority 2 ≤ priority 3. During DDS transmission, a QoS parameter optimization strategy is adopted, introducing a domain priority P, which is a normalized parameter with a value range of 0-1. Specifically, P = 0.9-1.0 for the vehicle safety domain, P = 0.6-0.8 for the vehicle comfort domain, and P = 0.3-0.5 for the vehicle entertainment domain. The vehicle safety domain includes functions related to collision warning and emergency braking; the vehicle comfort domain includes functions related to air conditioning adjustment and seat control; and the vehicle entertainment domain includes functions related to audio / video playback and navigation entertainment. The transmission delay control formula is as follows: in, The total transmission latency of data from the data radar to the regional cloud control center; This refers to the data transmission latency on the physical link, which includes the connection between the vehicle and the roadside, and between the roadside and the cloud control center. for The corresponding target state information dataset size, The compression ratio of the target state information; For the data packet processing delay between the data radar and the roadside terminal; This is the delay adjustment factor, which is dynamically determined based on the network congestion level.

5. The data radar-coordinated autonomous driving control method for vehicles based on the DDS protocol according to claim 1, characterized in that: In S4, the regional cloud control center uses HDFS to store target status information, MongoDB NoSQL database to store vehicle condition information, and Spark framework to implement parallel processing. Introducing vehicle health H using reinforcement learning algorithms: in, For the braking system weight, For battery weight, For tire weight, and ; This is a sub-item for braking system health, calculated based on brake pedal wear rate; This is a sub-item for battery health, calculated based on the remaining battery power. This is a sub-item for tire health, calculated based on tire pressure deviation. The reward function formula is: Where γ is the safety weight, Based on Calculate the collision risk coefficient based on the target location and trajectory. As an energy efficiency weight, Based on The target speed is used to adjust the vehicle's energy consumption strategy.

6. The data radar-coordinated autonomous driving control method for vehicles based on the DDS protocol according to claim 1, characterized in that: In S5, the data radar parsing unit is deployed on the vehicle-mounted MCU; The vehicle control unit includes: a power control submodule that controls the engine or motor speed, a brake control submodule that controls the brake master cylinder pressure, and a steering control submodule that controls the EPS steering motor. The control signals strictly comply with the CAN bus communication standard.

7. The data radar-coordinated autonomous driving control method for vehicles based on the DDS protocol according to claim 1, characterized in that: In S6, the update logic for topic information changes is as follows: first, a new topic carrying the updated information is created; then, the original data readers or writers are re-associated with the new topic; and finally, the original topic is deleted. The update logic for participant changes is as follows: first, create a new participant with updated parameters, then migrate the original publisher or subscriber and topic to the new participant, and finally cancel the original participant; QoS changes for data readers or writers only update the corresponding object, while domain changes require updating all associated participants, topics, publishers or subscribers, and data readers or writers.

8. The data radar-coordinated autonomous driving control method for vehicles based on the DDS protocol according to claim 1, characterized in that: It also includes S7: It uses a distributed caching unit to cache data readers, data writers and associated system numbers and service numbers, supports the management platform to update DCPS objects in real time, and enables messages to be transmitted along a specified path; The distributed caching unit uses a Redis cluster with a master-slave backup architecture, and is divided into independent cache partitions by domain; the cache update strategy process is as follows: ① Create a cached copy when an update request is received; ② Perform the update operation on the copy; ③ Switch cache access to the replica; ④ Delete old cached data.

9. The data radar-coordinated autonomous driving control method for vehicles based on the DDS protocol according to claim 1, characterized in that: It also includes S8: continuously sending link detection messages to the data radar, roadside signal receiving and transmitting terminal and regional cloud control center through the DDS adapter to verify the communication link status; if no detection response is received, the link is determined to be interrupted, triggering automatic switching from the main 5G link to the backup WIFI6 link to ensure data transmission continuity.

10. A data radar-coordinated autonomous driving control system for automobiles based on the DDS protocol, characterized in that: include: Management platform module: Deployed on a cloud server, including a configuration management sub-module that maintains the configuration information as described in claim 2 and supports visual editing on the web interface, and an API call sub-module that implements data radar model registration and configuration; Perception layer module: includes hardware unit and data fusion unit. The hardware unit includes one wide-angle camera on the windshield + two narrow-angle cameras in the rearview mirror, an on-board OBU supporting the C-V2XPC5 protocol and a roadside RSU supporting 5G+DSRC dual mode, one forward-facing lidar + four four-corner millimeter-wave radars. The data fusion unit is deployed on the data radar and executes the Kalman filtering algorithm described in claim 3. DDS Transmission Module: Includes a DDS adapter and a DCPS object unit. The DDS adapter integrates an ARM Cortex-A53 processor and has a built-in database query submodule for interfacing with MySQL and a DDS-REST protocol conversion submodule. The DCPS object unit is used to divide storage partitions by domain, QoS, and topic. Regional cloud control center module: deployed on an edge cloud server cluster, including a distributed storage unit and an intelligent decision-making unit, running the improved reinforcement learning algorithm described in claim 5; On-board execution module: includes a data radar analysis unit and a vehicle control unit. The data radar analysis unit is deployed on the on-board MCU. The vehicle control unit includes a power control submodule, a braking control submodule and a steering control submodule. Theme Management Module: Deployed on the roadside server, it includes a parameter monitoring submodule and a hierarchical update submodule. The parameter monitoring submodule is used to monitor DCPS object parameters, and the hierarchical update submodule is used to execute the update logic as described in claim 7. Cache module: adopts the Redis cluster architecture described in claim 8, including a partition management submodule and an update strategy submodule. The partition management submodule is used to divide partitions by domain, and the update strategy submodule is used to execute cache update strategies. Communication verification module: includes a link detection submodule and a link switching submodule, wherein the link detection submodule is used to send the detection message as described in claim 9, and the link switching submodule is used to execute the link switching logic as described in claim 9; Communication relationships between modules: The management platform communicates with the DDS transmission module via the REST protocol; the perception layer communicates with the DDS transmission module via the DDS protocol; the DDS transmission module communicates with the regional cloud control center via a 5G or fiber optic link; the regional cloud control center communicates with the vehicle execution module via the DDS protocol; and the topic management, caching, and communication verification modules communicate with the DDS transmission module via an internal bus, forming a closed loop throughout the entire process.

Images