Module vehicle transportation anti-collision monitoring system and method

Abstract

The invention discloses a module vehicle transportation anti-collision monitoring system and method. The system comprises an environment sensing module, a data preprocessing module, a coordinate and time processing module, a data fusion and association module, a target tracking and prediction module, an anti-collision decision module, a collision early warning and safe distance module and an early warning and interaction module. The method comprises; the method comprises the following steps: 1, environment perception and data acquisition; step 2, data preprocessing and coordinate time synchronization; step 3, target association and data fusion; step 4, target tracking and prediction; 5, performing anti-collision decision and path planning; step 6, collision early warning and safe distance management; according to the method, factors such as driver response time, vehicle braking performance and road adhesion conditions are comprehensively considered, a corrected safety distance threshold formula is established, and the accuracy of collision early warning is remarkably improved by combining TrickSim and Simulink joint simulation verification.

Description

Technical Field

[0001] This invention relates to the field of modular vehicle transportation technology, specifically to a modular vehicle transportation collision avoidance monitoring system and method. Background Technology

[0002] Modular vehicles are specialized equipment used for transporting and positioning large sections or steel structures. They typically have a strong load-bearing capacity, with most vehicles featuring a 6-axle design and a single vehicle capable of carrying a certain tonnage. The total load-bearing capacity can be increased by adding more vehicles, meeting the transportation needs of large sections. In the assembly process of large ships, modular vehicles play a crucial role, transporting large sections from the assembly site to the slipway and performing precise positioning and adjustments. This technology has significantly shortened the slipway cycle and improved production efficiency.

[0003] Existing modular vehicle transportation systems have the following drawbacks: First, the large mass and long braking distance of modular transport vehicles pose a significant challenge to traditional collision avoidance systems. In emergency braking or sudden situations, traditional systems often struggle to accurately and quickly assess collision risks, resulting in insufficient early warning and braking response, thus increasing the likelihood of accidents. This risk is particularly pronounced in complex and ever-changing transportation environments, placing extremely high demands on the driver's reaction speed and judgment. Secondly, the application of a single sensor in complex environments has limitations. Cameras are significantly affected by lighting conditions; in strong light, weak light, or dusty environments, the image acquisition quality of the camera will drop drastically, thus affecting the accuracy and reliability of target detection. Similarly, while millimeter-wave radar has certain advantages in medium- and long-range detection, its relatively low resolution makes it difficult to provide detailed environmental information, especially in close-range or complex obstacle recognition. Thirdly, instability in multi-target tracking is also a major challenge for modular vehicle transportation systems. During transportation, phenomena such as target occlusion, disappearance, and motion blur frequently occur. These phenomena severely interfere with the stability of tracking algorithms. Especially when targets undergo nonlinear motion, the performance of traditional tracking algorithms is often greatly limited, making it difficult to accurately track target trajectories and increasing safety hazards during transportation. Summary of the Invention

[0004] The purpose of this invention is to provide a collision avoidance monitoring system and method for modular vehicle transportation to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a modular vehicle transportation collision avoidance monitoring system, comprising an environmental perception module, a data preprocessing module connected to the environmental perception module, a coordinate and time processing module connected to the data preprocessing module, a data fusion and correlation module connected to the coordinate and time processing module, a target tracking and prediction module connected to the data fusion and correlation module, a collision avoidance decision module connected to the target tracking and prediction module, and a collision warning and safety distance module connected to the collision avoidance decision module.

[0006] As a further technical solution of the present invention, the collision warning and safe distance module includes a vehicle braking performance module, a road surface adhesion condition module, a safe distance threshold module, a simulation verification module, a collision warning module, and a data analysis module, and the environmental perception module includes a millimeter-wave radar module, a binocular camera module, a lidar scanning module, and an ultrasonic sensor module.

[0007] As a further technical solution of the present invention, the data preprocessing module includes a data receiving module, an empty target filtering module, a false target filtering module, and a non-dangerous target filtering module, and the coordinate and time processing module includes a coordinate transformation module and a time synchronization module.

[0008] As a further technical solution of the present invention, the data fusion and association module includes a global nearest neighbor algorithm module, a data fusion module, a stability evaluation module, and a tracking algorithm optimization module, and the target tracking and prediction module includes a motion information extraction module, an image motion fusion module, a Kalman filter module, a target tracking module, a target association module, and a tracking output module.

[0009] As a further technical solution of the present invention, the collision avoidance decision module includes a risk level assessment module, an emergency braking decision module, a path replanning module, and a blind spot monitoring module.

[0010] As a further technical solution of the present invention, the collision avoidance decision module is connected to a warning and interaction module, which includes an audio-visual alarm module, a HUD head-up display module, an in-vehicle voice prompt module, a remote monitoring module, a driving feedback module, and a vehicle control module.

[0011] A collision avoidance monitoring method for modular vehicle transportation includes the following steps: Step 1, environmental perception and data acquisition; Step 2, data preprocessing and coordinate time synchronization; Step 3, target association and data fusion; Step 4, target tracking and prediction; Step 5, collision avoidance decision-making and path planning; and Step 6, collision warning and safe distance management.

[0012] In step one above, environmental perception and data acquisition comprehensively acquire obstacle information through multi-sensor fusion technology.

[0013] In step two above, data preprocessing is synchronized with coordinate time to ensure data quality and spatiotemporal consistency.

[0014] In step three above, target association and data fusion are carried out to achieve effective integration and association of multi-sensor data;

[0015] In step four above, target tracking and prediction involves constructing a dynamic motion model and updating the target state in real time.

[0016] In step five above, collision avoidance decision-making and path planning make intelligent decisions based on the risk assessment results;

[0017] In step six above, collision warning and safe distance management ensure driving safety through the warning and interaction module.

[0018] As a further technical solution of the present invention, in step three, the Global Nearest Neighbor Algorithm Module in the data fusion and association module applies the GNN algorithm to realize target association between the binocular camera and the millimeter-wave radar. The data fusion module then fuses the data from the binocular camera and the millimeter-wave radar to improve the accuracy and robustness of target detection. The stability evaluation module then evaluates the stability of target tracking, identifies and handles problems such as occlusion, disappearance, and motion blur. Finally, the tracking algorithm optimization module optimizes the tracking algorithm based on the evaluation results to improve the accuracy and stability of target tracking.

[0019] As a further technical solution of the present invention, in step four, the motion information extraction module in the target tracking and prediction module uses the target's trajectory, velocity, and acceleration information to construct a dynamic motion model of the target. Then, the image motion fusion module fuses the target's motion information with the texture and color feature information in the image to improve the tracking accuracy and robustness. A convolutional neural network is used to extract image features, and combined with the target position predicted by the motion model, feature matching and tracking are performed. A weighted fusion strategy is used to balance the contributions of image information and motion information in the tracking process. The Kalman filter module uses the Kalman filter algorithm to predict and update the target state in real time. Then, the target tracking module implements the target tracking logic based on the target detection results and the Kalman filter prediction results. The target association module uses a deep learning model to learn the association features between targets and uses similarity measurement to determine whether the objects detected in different frames represent the same physical entity. Finally, the tracking output module outputs the target tracking results, including the target trajectory, velocity, and other information.

[0020] As a further technical solution of the present invention, in step six, the braking performance of the module vehicle, including braking deceleration and braking distance, is evaluated by the vehicle braking performance module in the collision warning and safe distance module to provide parameters for the safe distance model. The road surface adhesion condition module is used to monitor road surface conditions and adjust the parameters in the safe distance model to adapt to different road surface conditions. The safe distance threshold module is used to integrate various parameters and establish a safe distance threshold formula, which is:

[0021]

[0022] Where S is the safe distance threshold, V is the vehicle speed, T1 is the driver's reaction time, T2 is the braking time, T3 is the braking duration, and a max The maximum acceleration is given by ω, the slope correction factor is given by ω, the load factor is given by ω, and the safe stopping distance is given by d. A simulation verification module is then used to dynamically simulate the modular vehicle using TruckSim software, simulating driving states under different operating conditions. The simulation data from TruckSim is combined with Simulink to verify the effectiveness of the safe distance model. A collision warning module is used to determine whether the vehicle is at risk of collision based on the safe distance threshold and issue a warning signal. A data analysis module is used to collect real-vehicle test data, analyze algorithm performance, and optimize the safe distance model and collision warning algorithm.

[0023] Compared with existing technologies, the beneficial effects of this invention are as follows: By comprehensively considering factors such as driver reaction time, vehicle braking performance, and road surface adhesion conditions, this invention establishes a modified safe distance threshold formula and verifies it through joint simulation using TruckSim and Simulink, significantly improving the accuracy of collision warning. Furthermore, by employing environmental perception, it achieves comprehensive obstacle detection in the front and rear directions of the vehicle, enhancing perception capabilities in complex environments. By filtering out empty targets, false targets, and non-dangerous targets from millimeter-wave radar data, it improves data quality. Simultaneously, it achieves coordinate transformation and time synchronization between sensors, ensuring data consistency and accuracy. Moreover, it employs a global nearest neighbor algorithm to achieve target association and data fusion between sensors, replacing the target detection model with the EKF filtering algorithm, improving the tracking stability and accuracy of nonlinear moving targets. This significantly enhances the collision avoidance capability during modular vehicle transportation, providing a safer and more efficient solution for the transportation of large sections or steel structures. Attached Figure Description

[0024] Figure 1 This is a system structure diagram of the present invention;

[0025] Figure 2 This is a module architecture diagram of the coordinate and time processing module of the present invention;

[0026] Figure 3 This is a module architecture diagram of the target tracking and prediction module of the present invention;

[0027] Figure 4 This is a module architecture diagram of the collision avoidance decision module of the present invention;

[0028] Figure 5 This is a module architecture diagram of the collision warning and safety distance module of the present invention;

[0029] Figure 6 This is a system flowchart of the present invention;

[0030] Figure 7 This is a flowchart of the method of the present invention.

[0031] In the diagram: 1. Environmental perception module; 11. Millimeter-wave radar module; 12. Binocular camera module; 13. LiDAR scanning module; 14. Ultrasonic sensor module; 2. Data preprocessing module; 21. Data receiving module; 22. Empty target filtering module; 23. False target filtering module; 24. Non-dangerous target filtering module; 3. Coordinate and time processing module; 31. Coordinate transformation module; 32. Time synchronization module; 4. Data fusion and association module; 41. Global nearest neighbor algorithm module; 42. Data fusion module; 43. Stability evaluation module; 44. Tracking algorithm optimization module; 5. Target tracking and prediction module; 51. Motion information extraction module; 52. Image motion fusion module; 53. Kalman Array... Filtering module; 54. Target tracking module; 55. Target association module; 56. Tracking output module; 6. Collision avoidance decision module; 61. Risk level assessment module; 62. Emergency braking decision module; 63. Path replanning module; 64. Blind spot monitoring module; 7. Collision warning and safe distance module; 71. Vehicle braking performance module; 72. Road surface adhesion condition module; 73. Safe distance threshold module; 74. Simulation verification module; 75. Collision warning module; 76. Data analysis module; 8. Warning and interaction module; 81. Audible and visual alarm module; 82. HUD head-up display module; 83. In-vehicle voice prompt module; 84. Remote monitoring module; 85. Driving feedback module; 86. Vehicle control module. Detailed Implementation

[0032] 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.

[0033] Please see the appendix Figure 1 -Appendix Figure 6This invention provides an embodiment of a modular vehicle transportation collision avoidance monitoring system, comprising an environmental perception module 1, a data preprocessing module 2 connected to the environmental perception module 1, a coordinate and time processing module 3 connected to the data preprocessing module 2, a data fusion and correlation module 4 connected to the coordinate and time processing module 3, a target tracking and prediction module 5 connected to the data fusion and correlation module 4, a collision avoidance decision module 6 connected to the target tracking and prediction module 5, and a collision warning and safe distance module 7 connected to the collision avoidance decision module 6. The collision warning and safe distance module 7 includes a vehicle braking performance module 71, a road surface adhesion condition module 72, a safe distance threshold module 73, a simulation verification module 74, a collision warning module 75, and a data analysis module 76. The environmental perception module 1 includes a millimeter-wave radar module 11, a binocular camera module 12, a lidar scanning module 13, and an ultrasonic sensor module 14. The data preprocessing module 2 includes a data receiving module 21, an empty target filtering module 22, a false target filtering module 23, and a non-dangerous target filtering module 24. The coordinate and time processing module 3 includes a coordinate transformation module 31 and a time synchronization module 32. The data receiving module 21 is used to receive the raw data from the millimeter-wave radar, including information such as target position and speed. The empty target filtering module 22 is used to filter out empty targets in the millimeter-wave radar data to improve data quality. The false target filtering module 23 is used to identify and filter out false targets generated by noise or interference. The non-dangerous target filtering module 24 is used to filter out non-dangerous targets that do not pose a threat to vehicle driving according to the safe distance model. The coordinate transformation module 31 is used to realize the transformation between the millimeter-wave radar coordinate system, the world coordinate system, and the image coordinate system to ensure data spatial synchronization. The time synchronization module 32 uses a sampling period of 100ms to realize the time synchronization between the binocular camera and the millimeter-wave radar to ensure data time consistency.The data fusion and association module 4 includes a global nearest neighbor algorithm module 41, a data fusion module 42, a stability evaluation module 43, and a tracking algorithm optimization module 44. The target tracking and prediction module 5 includes a motion information extraction module 51, an image motion fusion module 52, a Kalman filter module 53, a target tracking module 54, a target association module 55, and a tracking output module 56. The global nearest neighbor algorithm module 41 applies the GNN algorithm to achieve target association between the binocular camera and the millimeter-wave radar. The data fusion module 42 fuses the data from the binocular camera and the millimeter-wave radar to improve the accuracy and robustness of target detection. Then, the stability evaluation module 43 evaluates the stability of target tracking, identifies and handles problems such as occlusion, disappearance, and motion blur. The tracking algorithm optimization module 44 optimizes the tracking algorithm based on the evaluation results to improve the accuracy and stability of target tracking. The motion information extraction module 51 uses the target's trajectory, velocity, and acceleration information to construct a dynamic motion model of the target. Then, the image motion fusion module 52 fuses the target's motion information with the texture and color features in the image to improve the tracking accuracy and robustness. A convolutional neural network is used. Image features are extracted and combined with the target position predicted by the motion model for feature matching and tracking. A weighted fusion strategy is used to balance the contributions of image information and motion information in the tracking process. The Kalman filter module 53 uses the Kalman filter algorithm to predict and update the target state in real time. Then, the target tracking module 54 implements the target tracking logic based on the target detection results and the Kalman filter prediction results. The target association module 55 uses a deep learning model to learn the association features between targets and uses similarity measurement to determine whether the objects detected in different frames represent the same physical entity. Finally, the target tracking result, including target trajectory, speed and other information, is output through the tracking output module 56. The collision avoidance decision module 6 includes a risk level assessment module 61, an emergency braking decision module 62, a path replanning module 63 and a blind spot monitoring module 64. The risk level assessment module 61 is used to calculate the collision risk level based on distance and speed. The emergency braking decision module 62 is used to trigger automatic braking or deceleration commands to avoid collision. The path replanning module 63 is used to generate obstacle avoidance paths in dangerous situations and link with the navigation system. The blind spot monitoring module 64 is used to monitor the blind spot of the rearview mirror and alert to lane change risks.The collision avoidance decision module 6 is connected to a warning and interaction module 8, which includes an audible and visual alarm module 81, a HUD head-up display module 82, an in-vehicle voice prompt module 83, a remote monitoring module 84, a driving feedback module 85, and a vehicle control module 86. The audible and visual alarm module 81 alerts the driver via a buzzer and flashing LEDs; the HUD head-up display module 82 projects warning information onto the windshield to reduce eye deflection; the in-vehicle voice prompt module 83 provides voice warnings; the remote monitoring module 84 sends real-time alerts and vehicle driving data to the cloud platform; the driving feedback module 85 monitors whether the warning is responded to and the effectiveness of the response; and the vehicle control module 86 controls the vehicle's movement.

[0034] Please see the appendix Figure 7 The present invention provides an embodiment of a collision avoidance monitoring method for modular vehicle transportation, comprising: step one, environmental perception and data acquisition; step two, data preprocessing and coordinate time synchronization; step three, target association and data fusion; step four, target tracking and prediction; step five, collision avoidance decision and path planning; and step six, collision warning and safe distance management.

[0035] In step one above, environmental perception and data acquisition comprehensively acquire obstacle information through multi-sensor fusion technology.

[0036] In step two above, data preprocessing is synchronized with coordinate time to ensure data quality and spatiotemporal consistency.

[0037] In step three above, target association and data fusion are used to effectively integrate and associate multi-sensor data. The global nearest neighbor algorithm module 41 in the data fusion and association module 4 applies the GNN algorithm to achieve target association between the binocular camera and the millimeter-wave radar. The data fusion module 42 fuses the data from the binocular camera and the millimeter-wave radar to improve the accuracy and robustness of target detection. Then, the stability evaluation module 43 evaluates the stability of target tracking, identifies and handles problems such as occlusion, disappearance, and motion blur. Based on the evaluation results, the tracking algorithm optimization module 44 optimizes the tracking algorithm to improve the accuracy and stability of target tracking.

[0038] In step four above, target tracking and prediction involves constructing a dynamic motion model and updating the target state in real time. The motion information extraction module 51 in the target tracking and prediction module 5 uses the target's trajectory, velocity, and acceleration information to construct a dynamic motion model. Then, the image motion fusion module 52 fuses the target's motion information with texture and color features in the image to improve tracking accuracy and robustness. A convolutional neural network is used to extract image features, which are then combined with the target position predicted by the motion model for feature matching and tracking. A weighted fusion strategy balances the contributions of image and motion information during the tracking process. The Kalman filter module 53 uses the Kalman filter algorithm to predict and update the target state in real time. The target tracking module 54 then implements the target tracking logic based on the target detection results and Kalman filter prediction results. The target association module 55 uses a deep learning model to learn the association features between targets and uses similarity metrics to determine whether objects detected in different frames represent the same physical entity. Finally, the tracking output module 56 outputs the target tracking results, including the target trajectory, velocity, and other information.

[0039] In step five above, collision avoidance decision-making and path planning make intelligent decisions based on the risk assessment results;

[0040] In step six above, collision warning and safe distance management ensures driving safety through the warning and interaction module 8. The vehicle braking performance module 71 within the collision warning and safe distance module 7 evaluates the vehicle's braking performance, including braking deceleration and braking distance, providing parameters for the safe distance model. The road surface adhesion condition module 72 monitors road conditions and adjusts the parameters in the safe distance model to adapt to different road conditions. The safe distance threshold module 73 integrates various parameters to establish a safe distance threshold formula, which is:

[0041]

[0042] Where S is the safe distance threshold, V is the vehicle speed, T1 is the driver's reaction time, T2 is the braking time, T3 is the braking duration, and a max The maximum acceleration is given by ω, the slope correction factor is given by ω, the load factor is given by ω, and the safe stopping distance is given by d. The simulation verification module 74 uses TruckSim software to perform dynamic simulations of the modular vehicle, simulating driving states under different working conditions. The simulation data from TruckSim is then combined with Simulink data to verify the effectiveness of the safe distance model. The collision warning module 75 determines whether the vehicle is at risk of collision based on the safe distance threshold and issues a warning signal. The data analysis module 76 collects real-vehicle test data, analyzes algorithm performance, and optimizes the safe distance model and collision warning algorithm.

[0043] Based on the above, the advantages of this invention are as follows: When using this invention for collision avoidance monitoring of modular vehicle transportation, firstly, environmental perception module 1 achieves environmental perception and data acquisition, comprehensively acquires obstacle information through multi-sensor fusion technology, and uses millimeter-wave radar module 11 for medium- and long-range obstacle detection, adapting to rain and snow weather and providing speed and distance data. Binocular camera module 12 is used to acquire image data, and lidar scanning module 13 generates a high-precision 3D environmental map through laser ranging, detecting obstacle outlines and distances. Ultrasonic sensor module 14 is used for near-range obstacle detection, suitable for low-speed scenarios such as parking assistance. Then, data preprocessing module 2 and coordinate and time processing module 3 perform data preprocessing and coordinate and time synchronization to ensure... To ensure data quality and spatiotemporal consistency, the system employs the following components: a data receiving module 21 receives raw data from the millimeter-wave radar, including target position and velocity information; an empty target filtering module 22 filters out empty targets from the millimeter-wave radar data to improve data quality; a false target filtering module 23 identifies and filters out false targets generated by noise or interference; a non-dangerous target filtering module 24 filters out non-dangerous targets that do not pose a threat to vehicle movement based on a safe distance model; a coordinate transformation module 31 performs conversions between the millimeter-wave radar coordinate system, the world coordinate system, and the image coordinate system to ensure data spatial synchronization; and a time synchronization module 32 uses a 100ms sampling period to achieve time synchronization between the binocular camera and the millimeter-wave radar, ensuring data temporal consistency. The system first performs target tracking, then uses the Global Nearest Neighbor Algorithm Module 41 in the Data Fusion and Association Module 4 to apply the GNN algorithm to achieve target association between the binocular camera and the millimeter-wave radar. The Data Fusion Module 42 then fuses the data from the binocular camera and the millimeter-wave radar to improve the accuracy and robustness of target detection. The Stability Evaluation Module 43 then evaluates the stability of target tracking, identifying and handling issues such as occlusion, disappearance, and motion blur. The Tracking Algorithm Optimization Module 44 optimizes the tracking algorithm based on the evaluation results, improving the accuracy and stability of target tracking. Finally, target tracking and prediction are performed, constructing a dynamic motion model and updating the target state in real time. The Motion Information Extraction Module 51 in the Target Tracking and Prediction Module 5 utilizes the target's motion... The system collects trajectory, velocity, and acceleration information to construct a dynamic motion model of the target. Then, the image motion fusion module 52 fuses the target's motion information with texture and color features from the image, improving tracking accuracy and robustness. A convolutional neural network extracts image features and combines them with the target position predicted by the motion model for feature matching and tracking. A weighted fusion strategy balances the contributions of image and motion information during tracking. The Kalman filter module 53 uses the Kalman filter algorithm to predict and update the target state in real time. The target tracking module 54 then implements the target tracking logic based on the target detection results and Kalman filter prediction results. Finally, the target association module 55 uses a deep learning model to learn the association features between targets.Similarity measurement determines whether objects detected in different frames represent the same physical entity. The tracking output module 56 outputs the target tracking results, including target trajectory, speed, and other information. The collision avoidance decision module 6 performs collision avoidance decisions and path planning, making intelligent decisions based on risk assessment results. The risk level assessment module 61 calculates the collision risk level based on distance and speed. The emergency braking decision module 62 triggers automatic braking or deceleration commands to avoid a collision. The path replanning module 63 generates obstacle avoidance paths in dangerous situations and links with the navigation system. The blind spot monitoring module 64 monitors rearview mirror blind spots, alerts to lane change risks, and provides warnings through the warning and interaction module 8. The audible and visual alarm module 81 alerts the driver via a buzzer and LED flashing. The HUD head-up display module 82... Warning information is projected onto the windshield to reduce line-of-sight deviation. The in-vehicle voice prompt module 83 provides voice warnings. The remote monitoring module 84 sends real-time alerts and vehicle driving data to the cloud platform. The driving feedback module 85 monitors whether the warnings are responded to and the effectiveness of the response. The vehicle control module 86 controls the vehicle's movement. Driving safety is ensured through the warning and interaction module 8. The vehicle braking performance module 71 within the collision warning and safe distance module 7 evaluates the vehicle's braking performance, including braking deceleration and braking distance, providing parameters for the safe distance model. The road surface adhesion condition module 72 monitors road conditions and adjusts parameters in the safe distance model to adapt to different road conditions. The safe distance threshold module 73 integrates various parameters to establish a safe distance threshold formula: [Formula omitted].

[0044]

[0045] Where S is the safe distance threshold, V is the vehicle speed, T1 is the driver's reaction time, T2 is the braking time, T3 is the braking duration, and a max The maximum acceleration is given by ω, the slope correction factor is given by ω, the load factor is given by ω, and the safe stopping distance is given by d. The simulation verification module 74 uses TruckSim software to perform dynamic simulations of the modular vehicle, simulating driving states under different working conditions. The simulation data from TruckSim is then combined with Simulink data to verify the effectiveness of the safe distance model. The collision warning module 75 determines whether the vehicle is at risk of collision based on the safe distance threshold and issues a warning signal. The data analysis module 76 collects real-vehicle test data, analyzes algorithm performance, and optimizes the safe distance model and collision warning algorithm.

[0046] 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 modular vehicle transportation collision avoidance monitoring system, comprising an environmental perception module (1), characterized in that: The environmental perception module (1) is connected to a data preprocessing module (2), the data preprocessing module (2) is connected to a coordinate and time processing module (3), the coordinate and time processing module (3) is connected to a data fusion and association module (4), the data fusion and association module (4) is connected to a target tracking and prediction module (5), the target tracking and prediction module (5) is connected to a collision avoidance decision module (6), and the collision avoidance decision module (6) is connected to a collision warning and safe distance module (7).

2. The modular vehicle transportation collision avoidance monitoring system according to claim 1, characterized in that: The collision warning and safe distance module (7) includes a vehicle braking performance module (71), a road surface adhesion condition module (72), a safe distance threshold module (73), a simulation verification module (74), a collision warning module (75), and a data analysis module (76). The environmental perception module (1) includes a millimeter-wave radar module (11), a binocular camera module (12), a lidar scanning module (13), and an ultrasonic sensor module (14).

3. The modular vehicle transportation collision avoidance monitoring system according to claim 1, characterized in that: The data preprocessing module (2) includes a data receiving module (21), an empty target filtering module (22), a false target filtering module (23), and a non-dangerous target filtering module (24). The coordinate and time processing module (3) includes a coordinate transformation module (31) and a time synchronization module (32).

4. The modular vehicle transport collision avoidance monitoring system according to claim 1, characterized in that: The data fusion and association module (4) includes a global nearest neighbor algorithm module (41), a data fusion module (42), a stability evaluation module (43), and a tracking algorithm optimization module (44). The target tracking and prediction module (5) includes a motion information extraction module (51), an image motion fusion module (52), a Kalman filter module (53), a target tracking module (54), a target association module (55), and a tracking output module (56).

5. The modular vehicle transport collision avoidance monitoring system according to claim 1, characterized in that: The collision avoidance decision module (6) includes a risk level assessment module (61), an emergency braking decision module (62), a path replanning module (63), and a blind spot monitoring module (64).

6. The modular vehicle transport collision avoidance monitoring system according to claim 1, characterized in that: The collision avoidance decision module (6) is connected to an early warning and interaction module (8), which includes an audible and visual alarm module (81), a HUD head-up display module (82), an in-vehicle voice prompt module (83), a remote monitoring module (84), a driving feedback module (85), and a vehicle control module (86).

7. A collision avoidance monitoring method for modular vehicle transportation, comprising: Step 1, environmental perception and data acquisition; Step 2, data preprocessing and coordinate time synchronization; Step 3, target association and data fusion; Step 4, target tracking and prediction; Step 5, collision avoidance decision-making and path planning; Step 6, collision warning and safe distance management; characterized in that: In step one above, environmental perception and data acquisition comprehensively acquire obstacle information through multi-sensor fusion technology. In step two above, data preprocessing is synchronized with coordinate time to ensure data quality and spatiotemporal consistency. In step three above, target association and data fusion are carried out to achieve effective integration and association of multi-sensor data; In step four above, target tracking and prediction involves constructing a dynamic motion model and updating the target state in real time. In step five above, collision avoidance decision-making and path planning make intelligent decisions based on the risk assessment results; In step six above, collision warning and safe distance management ensure driving safety through the warning and interaction module (8).

8. A method for preventing collisions during modular vehicle transportation according to claim 7, characterized in that: In step three, the GNN algorithm is applied through the global nearest neighbor algorithm module (41) in the data fusion and association module (4) to realize the target association between the binocular camera and the millimeter-wave radar. The data of the binocular camera and the millimeter-wave radar are fused through the data fusion module (42) to improve the accuracy and robustness of target detection. Then, the stability evaluation module (43) evaluates the stability of target tracking, identifies and handles problems such as occlusion, disappearance and motion blur, and optimizes the tracking algorithm based on the evaluation results through the tracking algorithm optimization module (44) to improve the accuracy and stability of target tracking.

9. A method for monitoring collision avoidance during modular vehicle transportation according to claim 7, characterized in that: In step four, the motion information extraction module (51) in the target tracking and prediction module (5) uses the target's motion trajectory, speed and acceleration information to construct a dynamic motion model of the target. Then, the motion information of the target is fused with the texture and color feature information in the image through the image motion fusion module (52) to improve the tracking accuracy and robustness. The image features are extracted by using a convolutional neural network and combined with the target position predicted by the motion model to perform feature matching and tracking. The contribution of image information and motion information in the tracking process is balanced by a weighted fusion strategy. The Kalman filter module (53) uses the Kalman filter algorithm to predict and update the target state in real time. Then, the target tracking module (54) implements the target tracking logic based on the target detection results and the Kalman filter prediction results. The target association module (55) uses a deep learning model to learn the association features between targets and determines whether the objects detected between different frames represent the same physical entity through similarity measurement. Finally, the target tracking results, including target trajectory, speed and other information, are output through the tracking output module (56).

10. A method for preventing collisions during modular vehicle transportation according to claim 7, characterized in that: In step six, the braking performance of the module vehicle, including braking deceleration and braking distance, is evaluated by the vehicle braking performance module (71) in the collision warning and safe distance module (7), providing parameters for the safe distance model. The road surface adhesion condition module (72) is used to monitor road surface conditions and adjust the parameters in the safe distance model to adapt to different road surface conditions. The safe distance threshold module (73) is used to integrate various parameters and establish a safe distance threshold formula, which is: Where S is the safe distance threshold, V is the vehicle speed, T1 is the driver's reaction time, T2 is the braking time, T3 is the braking duration, and a max The maximum acceleration is w, the slope correction coefficient is l, the load coefficient is d, and the safe stopping distance is d. Then, the simulation verification module (74) is used to perform dynamic simulation of the modular vehicle using TruckSim software, simulate the driving state under different working conditions, and combine the simulation data of TruckSim with Simulink to verify the effectiveness of the safe distance model. The collision warning module (75) is used to determine whether the vehicle is in collision risk according to the safe distance threshold and issue a warning signal. The data analysis module (76) is used to collect real vehicle test data, analyze algorithm performance, and optimize the safe distance model and collision warning algorithm.

Images