Unmanned vehicle formation trajectory planning control system and method

Abstract

The present invention belongs to the field of automatic driving, and discloses a control system and method for unmanned vehicle formation trajectory planning. The present invention regards the vehicle formation problem as a problem of position change scheme of slave vehicles based on the master vehicle. The present invention can realize flexible formation under the premise of safe and fast driving, that is, safe obstacle avoidance in off-road environment. The master vehicle is implanted with a trained formation trajectory control model to command the slave vehicles according to the surrounding environment and the number and status of the slave vehicles, thereby realizing automatic formation change. The model obtains a safety value judgment based on the processing of the surrounding environment and the number and status of the slave vehicles, and determines whether to join the model for model training and updating according to the safety value. The present invention identifies the accuracy of formation trajectory control through the safe self-training update of the model, only through the environmental detection of the master vehicle combined with the status of the slave vehicles, and increases the safety value constraint, thereby improving the generalization ability of the model. The unmanned vehicle formation can obtain safety guarantees, flexibly change the formation, and reduce the control risk of the team driving.

Description

Technical Field

[0001] The present invention belongs to the technical field of autonomous driving, and in particular relates to a control system and method for trajectory planning of an unmanned vehicle formation. Background Art

[0002] Autonomous vehicles have a research history of decades. They can replace humans to complete tedious tasks in complex scenarios such as high-density, long-cycle, and high-traffic traffic, and have high social and economic value. Highways have the characteristics of clear topological structure, known traffic rules, clear restrictions, and relative closure, which are typical scenarios for the implementation of autonomous driving. However, there are still big problems with formation tasks on off-road roads. First, the dynamic vehicle motion state on off-road roads is complex, and vehicle formation coordination is difficult; second, vehicle perception is limited, and the stability of the formation system is poor; third, the fixed formation mode makes the system inflexible and has a greater impact on surrounding vehicles. Faced with complex and changeable off-road road scenes, fixed control modes will also lose system flexibility and adaptability to environmental changes.

[0003] At present, the known unmanned vehicle collaborative formation planning and control methods mainly include the following three methods: pilot-follow method, behavior-based method and virtual structure method; among them, although the pilot-follow method has strong scalability and is simple and easy to implement, the robustness of the formation is highly dependent on the robustness of the pilot, and it is difficult to achieve the expected effect; although the behavior-based method can quickly draw control conclusions in complex environments, it is uncertain whether there is incompatibility when integrating various behaviors, and the integrated method is difficult to meet various conditions; the advantage of the virtual structure method is that the tracking trajectory is highly accurate and easy to observe, but the formation is single and it is difficult to apply to more scenarios.

[0004] Reinforcement learning is a type of machine learning. With the development of artificial intelligence and machine learning, reinforcement learning has gradually been applied to autonomous driving tasks. Reinforcement learning does not rely on pre-labeled data sets. Although it has strong generalization capabilities and can more effectively solve special situations in the environment, the construction of reinforcement learning models is relatively complex and cannot be well applied to off-road scenarios.

[0005] Therefore, the existing unmanned vehicle collaborative formation planning and control methods have poor robustness and cannot be applied to the unmanned vehicle formation trajectory planning and control in off-road scenarios. The trajectory planning and control effect in complex off-road scenarios is poor.

[0006] The above contents are only used to assist in understanding the technical solution of the present invention and do not constitute an admission that the above contents are prior art.

[0007] Therefore, there is an urgent need for an unmanned vehicle formation trajectory planning control system and method to solve the above problems. Summary of the invention

[0008] In order to solve the above technical problems, the present invention proposes a trajectory planning control system and method for an unmanned vehicle formation.

[0009] In a first aspect of the present invention, a method for trajectory planning and control of an unmanned vehicle formation is provided, wherein the unmanned vehicle formation comprises a master vehicle and a slave vehicle, and the master vehicle and the slave vehicle communicate with each other through a network, wherein the method comprises the following steps:

[0010] P1. The main vehicle obtains the first drivable channel information of the drivable channel between the obstacles on both sides of the road, obtains the first follower vehicle status information of other follower vehicles in the unmanned vehicle formation, and obtains the first trajectory planning formation that can pass through the drivable channel at this time;

[0011] P2: The main vehicle constructs a first formation trajectory control model according to the first drivable channel information, the first slave vehicle state information and the first trajectory planning formation at this time;

[0012] P3, the main vehicle receives the second drivable channel information of the drivable channel between the obstacles on both sides of the road, and obtains the second slave vehicle status information of other slave vehicles in the unmanned vehicle formation, the formation trajectory control model processes the second drivable channel information and the second slave vehicle status information to generate a second trajectory planning formation, and the main vehicle controls the entire vehicle fleet to drive in formation according to the second trajectory planning formation;

[0013] P4, obtaining a first safety value according to the second drivable channel information and the second slave vehicle status information, and comparing the first safety value with a set safety threshold to determine whether to join the training of the second formation trajectory control model;

[0014] P5. A first safety value is obtained according to the second drivable channel information and the second slave vehicle status information, and the first safety value is compared with a set safety threshold to determine whether to add it to the training of the second formation trajectory control model. If the first safety value is greater than the safety threshold, a second formation trajectory control model is constructed by using the first drivable channel information, the first slave vehicle status information, the second drivable channel information, and the second slave vehicle status information. The main vehicle updates the first formation trajectory control model to the newly constructed second formation trajectory control model, and performs subsequent trajectory planning formation formation according to the second formation trajectory control model.

[0015] Furthermore, the first drivable channel information or the second drivable channel information includes the width d between the obstacle objects on both sides of the drivable channel and the angle between the host vehicle and the obstacle objects on both sides detected in real time. , the length L of the drivable channel formed by obstacles on both sides is obtained by processing the distance and angle between the obstacle objects in the drivable channel and the main vehicle obtained by the laser radar of the main vehicle.

[0016] Furthermore, the first slave vehicle status information or the second slave vehicle status information is the position information of the slave vehicle relative to the main vehicle in a coordinate system constructed with the main vehicle as the coordinate origin, as well as the vehicle speed and driving direction and the number of slave vehicles.

[0017] Furthermore, the coordinate system is a Cartesian coordinate system or a Frenet coordinate system.

[0018] Furthermore, the first safety value is obtained by processing the second drivable channel information and the second slave vehicle status information.

[0019] Furthermore, the first formation trajectory control model or the second formation trajectory control model adopts an improved model based on deep learning.

[0020] Furthermore, the first formation trajectory control model or the second formation trajectory control model adopts improved deep learning.

[0021] A unmanned vehicle formation trajectory planning and control system is also provided, the system comprising a master vehicle data acquisition module, a slave vehicle status data acquisition module, a master vehicle data processing module, a formation trajectory control model building module, a formation trajectory control model updating module, and a master vehicle formation trajectory control module, characterized in that:

[0022] The main vehicle data acquisition module includes a main vehicle laser radar, which is used to obtain first drivable channel information of the drivable channel between obstacles on both sides of the road, and is also used to obtain second drivable channel information of the drivable channel between obstacles on both sides of the road, and construct a coordinate system according to the main vehicle, and also obtain the first trajectory planning formation at this time;

[0023] The slave vehicle status data acquisition module is wirelessly connected to the master vehicle data processing module to obtain first slave vehicle status information between the slave vehicle and the master vehicle, and also obtain second slave vehicle status information between the slave vehicle and the master vehicle;

[0024] The master vehicle data processing module is configured to obtain a first safety value according to the second drivable channel information and the second slave vehicle state information;

[0025] The formation trajectory control model construction module: constructs a first formation trajectory control model according to the first drivable channel information, the first slave vehicle state information and the first trajectory planning formation, and also constructs a second formation trajectory control model according to the first drivable channel information, the first slave vehicle state information combined with the second drivable channel information and the second slave vehicle state information;

[0026] The formation trajectory control model updating module: obtains a first safety value according to the second drivable channel information and the second slave vehicle state information, and compares the first safety value with a set safety threshold to determine whether to add to the training of the second formation trajectory control model;

[0027] The main vehicle formation trajectory control module: calls the first formation trajectory control model to process the second drivable channel information and the second slave vehicle status information to generate a second trajectory planning formation, and calls the second formation trajectory control model to process the subsequently obtained drivable channel information and slave vehicle status information to generate a real-time trajectory planning formation, and controls the driving formation of the unmanned vehicle formation according to the real-time trajectory planning formation.

[0028] Furthermore, the first drivable channel information or the second drivable channel information includes the width d between the obstacle objects on both sides of the drivable channel and the angle between the host vehicle and the obstacle objects on both sides detected in real time. The length L and width d of the drivable channel formed by the obstacles on both sides are obtained by the distance and angle between the obstacle objects in the drivable channel and the main vehicle obtained by the laser radar of the main vehicle.

[0029] Furthermore, the first formation trajectory control model or the second formation trajectory control model adopts an improved model based on deep learning;

[0030] The first formation trajectory control model or the second formation trajectory control model adopts improved deep learning.

[0031] The vehicle formation problem is regarded as a problem of position change plan of slave vehicles based on the main vehicle. The present invention can realize flexible formation under the premise of safe and fast driving, that is, safe obstacle avoidance in off-road environment. The main vehicle is implanted with a trained formation trajectory control model to command the slave vehicles according to the surrounding environment and the number and status of slave vehicles, so as to realize automatic formation change. The model obtains safety value judgment according to the processing of the surrounding environment and the number and status of slave vehicles, and determines whether to join the model for model training and updating according to the safety value. The present invention improves the safe self-training update of the neural network model using off-road scenarios, and identifies the accuracy of formation trajectory control only through the environmental detection of the main vehicle combined with the status of the slave vehicles, increases the safety value constraint, and thus improves the generalization ability of the model. The unmanned vehicle formation can obtain safety guarantees, flexibly change the formation, and reduce the control risk of the fleet driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flow chart of a trajectory planning and control method of an unmanned vehicle formation of the present invention;

[0033] Figure 2It is a structural schematic diagram of a trajectory planning control system for an unmanned vehicle formation of the present invention;

[0034] Figure 3 is a schematic diagram of the Frenet coordinate system in the present invention;

[0035] Figure 4 It is a schematic diagram of the principle of the neural network model in the present invention;

[0036] Figure 5 It is the driving trajectory diagram of the fleet in the present invention;

[0037] Figure 6 This is a schematic diagram of the vehicle fleet's driving trajectory change in the present invention; DETAILED DESCRIPTION

[0038] The invention is further described below in conjunction with the accompanying drawings and specific implementation methods.

[0039] The first embodiment of the present invention:

[0040] In a first aspect of the present invention, in order to solve the above-mentioned technical problems, a control system and method for trajectory planning of an unmanned vehicle formation is provided.

[0041] In a first aspect of the present invention, a method for trajectory planning and control of an unmanned vehicle formation is provided, wherein the unmanned vehicle formation comprises a master vehicle and a slave vehicle, and the master vehicle and the slave vehicle communicate with each other through a network, wherein the method comprises the following steps:

[0042] P1. The main vehicle obtains the first drivable channel information of the drivable channel between the obstacles on both sides of the road, obtains the first follower vehicle status information of other follower vehicles in the unmanned vehicle formation, and obtains the first trajectory planning formation that can pass through the drivable channel at this time;

[0043] P2: The main vehicle constructs a first formation trajectory control model according to the first drivable channel information, the first slave vehicle state information and the first trajectory planning formation at this time;

[0044] P3, the main vehicle receives the second drivable channel information of the drivable channel between the obstacles on both sides of the road, and obtains the second slave vehicle status information of other slave vehicles in the unmanned vehicle formation, the first formation trajectory control model processes the second drivable channel information and the second slave vehicle status information to generate a second trajectory planning formation, and the main vehicle controls the entire vehicle fleet to drive in formation according to the second trajectory planning formation;

[0045] P4, obtaining a first safety value according to the second drivable channel information and the second slave vehicle status information, and comparing the first safety value with a set safety threshold to determine whether to join the training of the second formation trajectory control model;

[0046] P5. A first safety value is obtained according to the second drivable channel information and the second slave vehicle status information, and the first safety value is compared with a set safety threshold to determine whether to add it to the training of the second formation trajectory control model. If the first safety value is greater than the safety threshold, a second formation trajectory control model is constructed by using the first drivable channel information, the first slave vehicle status information, the second drivable channel information, and the second slave vehicle status information. The main vehicle updates the first formation trajectory control model to the newly constructed second formation trajectory control model, and performs subsequent trajectory planning formation formation according to the second formation trajectory control model.

[0047] Furthermore, the first drivable channel information or the second drivable channel information includes the width d between the obstacle objects on both sides of the drivable channel and the angle between the host vehicle and the obstacle objects on both sides detected in real time. , the length L of the drivable channel formed by obstacles on both sides is obtained by processing the obstacle objects in the drivable channel obtained by the laser radar of the main vehicle. The specific calculation formula is:

[0048]

[0049] In the formula is the correction factor obtained according to the obstacle type, is the distance between the main vehicle and the obstacle on one side measured by the laser radar. It is the distance between the main vehicle and the obstacle on the other side measured by the laser radar.

[0050] Furthermore, the first slave vehicle status information or the second slave vehicle status information is the position information of the slave vehicle relative to the master vehicle in a coordinate system constructed with the master vehicle as the coordinate origin, as well as the vehicle speed and driving direction and the number of slave vehicles.

[0051] Furthermore, the coordinate system is a Cartesian coordinate system or a Frenet coordinate system.

[0052] Furthermore, the first safety value is obtained according to the second drivable channel information and the second slave vehicle state information, and the specific calculation formula of the first safety value s is:

[0053]

[0054] Where n is the number of slave vehicles, , ,..., represents the speed of vehicle 1, vehicle 2, ..., vehicle n, t is the human reaction time, and k is the safety adjustment factor set according to the off-road environment.

[0055] In this embodiment, the reaction time t is 0.5s, k is a value automatically selected according to the road conditions visually recognized by the main vehicle, if the road is a muddy road, k is 0.01-0.02, if the road is a cement road, k is 0.05-0.08, and the value of n is greater than or equal to 2.

[0056] Furthermore, the first formation trajectory control model or the second formation trajectory control model adopts an improved model based on deep learning.

[0057] Furthermore, the first formation trajectory control model or the second formation trajectory control model adopts improved deep learning, and its activation function As shown below:

[0058]

[0059] is the activation function value, X is the drivable channel information and is obtained from the vehicle status information. is the correction factor obtained according to the obstacle type, k is the safety adjustment factor set according to the off-road environment, and n is the number of follower vehicles.

[0060] In this embodiment is the correction factor obtained according to the obstacle type. If the road is a muddy road, then The value range is 0.1-0.5. If the road is a cement road, The value range is 0.6-0.9.

[0061] In this embodiment, the fleet formation change follows the following process. First, the automatic formation system of unmanned vehicles is mainly suitable for multi-vehicle collaborative task scenarios on unstructured roads in off-road environments.

[0062] The algorithm for platoon driving adopts a formation control method based on the fusion of position information and perception information.

[0063] The general process is:

[0064] Initial information collection: The main vehicle collects perception information of the current environment and the number of vehicles performing the task together;

[0065] Formation judgment: Analyze and determine the most efficient formation in the current environment (in order to facilitate the task, give priority to the regular triangle or inverted triangle formation; when driving on narrow roads, change to a vertical formation based on the information of the drivable channel; under specific task requirements, use a horizontal formation);

[0066] Slave vehicle positioning: Based on the current expected driving formation, each vehicle determines its relative position in the formation;

[0067] Formation formation: In the initial stage, the main vehicle drives along the navigation trajectory, and the follower vehicles drive to the predetermined positions in the formation in sequence (using the formation trajectory control model model predictive control to ensure stable formation and spacing);

[0068] Formation driving: After the lead vehicle confirms that the formation is formed, it speeds up;

[0069] Single vehicle departure: When a task vehicle needs to leave the formation, the task vehicle predicts the trajectory of other vehicles and drives to the task point by avoiding obstacles. At this time, the data from the number of vehicles is updated again based on the model training.

[0070] Formation recovery: After the task vehicle completes the task, it returns to the formation according to the relative position specified in the slave vehicle positioning stage and continues to drive. At this time, whether the slave vehicle magnitude data is re-updated with the model training;

[0071] Automatic formation change: The main vehicle continuously collects information and analyzes the current environment to determine whether a formation change is necessary;

[0072] Repeat the "slave car positioning" stage to redefine the relative position of each car in the formation.

[0073] A unmanned vehicle formation trajectory planning and control system is also provided, the system comprising a master vehicle data acquisition module, a slave vehicle status data acquisition module, a master vehicle data processing module, a formation trajectory control model building module, a formation trajectory control model updating module, and a master vehicle formation trajectory control module, characterized in that:

[0074] The main vehicle data acquisition module includes a main vehicle laser radar, which is used to obtain first drivable channel information of the drivable channel between obstacles on both sides of the road, and is also used to obtain second drivable channel information of the drivable channel between obstacles on both sides of the road, and construct a coordinate system according to the main vehicle, and also obtain the first trajectory planning formation at this time;

[0075] The slave vehicle status data acquisition module is wirelessly connected to the master vehicle data processing module to obtain first slave vehicle status information between the slave vehicle and the master vehicle, and also obtain second slave vehicle status information between the slave vehicle and the master vehicle;

[0076] The master vehicle data processing module is configured to obtain a first safety value according to the second drivable channel information and the second slave vehicle state information;

[0077] The formation trajectory control model construction module: constructs a first formation trajectory control model according to the first drivable channel information, the first slave vehicle state information and the first trajectory planning formation, and also constructs a second formation trajectory control model according to the first drivable channel information, the first slave vehicle state information combined with the second drivable channel information and the second slave vehicle state information;

[0078] The formation trajectory control model updating module: obtains a first safety value according to the second drivable channel information and the second slave vehicle state information, and compares the first safety value with a set safety threshold to determine whether to add to the training of the second formation trajectory control model;

[0079] The main vehicle formation trajectory control module: calls the first formation trajectory control model to process the second drivable channel information and the second slave vehicle status information to generate a second trajectory planning formation, and calls the second formation trajectory control model to process the subsequently obtained drivable channel information and slave vehicle status information to generate a real-time trajectory planning formation, and controls the driving formation of the unmanned vehicle formation according to the real-time trajectory planning formation.

[0080] Furthermore, the first drivable channel information or the second drivable channel information includes the width d between the obstacle objects on both sides of the drivable channel and the angle between the host vehicle and the obstacle objects on both sides detected in real time. , the length L of the drivable channel formed by obstacles on both sides is obtained by processing the obstacle objects in the drivable channel obtained by the laser radar of the main vehicle. The specific calculation formula is:

[0081]

[0082] In the formula is the correction factor obtained according to the obstacle type, is the distance between the main vehicle and the obstacle on one side measured by the laser radar. It is the distance between the main vehicle and the obstacle on the other side measured by the laser radar.

[0083] Furthermore, the first formation trajectory control model or the second formation trajectory control model adopts an improved model based on deep learning;

[0084] The first formation trajectory control model or the second formation trajectory control model adopts improved deep learning, and its activation function As shown below:

[0085]

[0086] is the activation function value, X is the drivable channel information and is obtained from the vehicle status information. is the correction factor obtained according to the obstacle type, k is the safety adjustment factor set according to the off-road environment, and n is the number of follower vehicles.

[0087] The vehicle formation problem is regarded as a problem of position change plan of slave vehicles based on the main vehicle. The present invention can realize flexible formation under the premise of safe and fast driving, that is, safe obstacle avoidance in off-road environment. The main vehicle is implanted with a trained formation trajectory control model to command the slave vehicles according to the surrounding environment and the number and status of slave vehicles, so as to realize automatic formation change. The model obtains safety value judgment according to the processing of the surrounding environment and the number and status of slave vehicles, and determines whether to join the model for model training and updating according to the safety value. The present invention improves the safe self-training update of the neural network model using off-road scenarios, and identifies the accuracy of formation trajectory control only through the environmental detection of the main vehicle combined with the status of the slave vehicles, increases the safety value constraint, and thus improves the generalization ability of the model. The unmanned vehicle formation can obtain safety guarantees, flexibly change the formation, and reduce the control risk of the fleet driving.

[0088] A combination of multiple embodiments of the present invention can achieve all of the above effects, but it is not required that each embodiment of the present invention achieve all of the above advantages and effects, because each embodiment of the present invention can constitute an independent technical solution and make one or more contributions to the prior art.

[0089] The present invention does not particularly specify the structure of some modules, which shall be subject to the contents recorded in the prior art. The prior art mentioned in the above background technology section and the specific embodiment section of the present invention can be used as part of the present invention to understand the meaning of some technical features or parameters. The scope of protection of the present invention shall be subject to the contents actually recorded in the claims.

Claims

1. A trajectory planning and control method for an unmanned vehicle formation, wherein the unmanned vehicle formation includes a master vehicle and a slave vehicle, and the master vehicle and the slave vehicle communicate with each other through a network, characterized in that: The method comprises the following steps: P1. The main vehicle obtains the first drivable channel information of the drivable channel between the obstacles on both sides of the road, obtains the first follower vehicle status information of other follower vehicles in the unmanned vehicle formation, and obtains the first trajectory planning formation that can pass through the drivable channel at this time; P2: The main vehicle constructs a first formation trajectory control model according to the first drivable channel information, the first slave vehicle state information and the first trajectory planning formation at this time; P3, the main vehicle receives the second drivable channel information of the drivable channel between the obstacles on both sides of the road, and obtains the second slave vehicle status information of other slave vehicles in the unmanned vehicle formation, the first formation trajectory control model processes the second drivable channel information and the second slave vehicle status information to generate a second trajectory planning formation, and the main vehicle controls the entire vehicle fleet to drive in formation according to the second trajectory planning formation; P4, obtaining a first safety value according to the second drivable channel information and the second slave vehicle status information, and comparing the first safety value with a set safety threshold to determine whether to join the training of the second formation trajectory control model; P5. Obtain a first safety value according to the second drivable channel information and the second slave vehicle status information, and compare the first safety value with a set safety threshold to determine whether to join the training of the second formation trajectory control model. If the first safety value is greater than the safety threshold, construct a second formation trajectory control model by using the first drivable channel information, the first slave vehicle status information, the second drivable channel information, and the second slave vehicle status information. The master vehicle updates the first formation trajectory control model to the newly constructed second formation trajectory control model, and performs subsequent trajectory planning formation formation according to the second formation trajectory control model. The first drivable channel information or the second drivable channel information includes the width d between the obstacles on both sides of the drivable channel and the angle between the host vehicle and the obstacles on both sides detected in real time. The length L of the drivable channel formed by the obstacles on both sides and the width d are obtained by processing the distance and angle between the obstacle objects in the drivable channel and the main vehicle obtained by the laser radar of the main vehicle. The specific calculation formula is: ； In the formula is the correction factor obtained according to the obstacle type, is the distance between the main vehicle and the obstacle on one side measured by the laser radar. The distance between the main vehicle and the obstacle on the other side measured by the laser radar; The first slave vehicle status information or the second slave vehicle status information is the position information of the slave vehicle relative to the main vehicle in a coordinate system constructed with the main vehicle as the coordinate origin, as well as the vehicle speed and driving direction and the number of slave vehicles.

2. The unmanned vehicle formation trajectory planning and control method according to claim 1, characterized in that: The coordinate system is a Cartesian coordinate system or a Frenet coordinate system.

3. The unmanned vehicle formation trajectory planning and control method according to claim 2, characterized in that: The first safety value is obtained by processing the second drivable channel information and the second slave vehicle state information. The specific calculation formula of the first safety value s is: ； Where n is the number of slave vehicles, , ,..., represents the speed of vehicle 1, vehicle 2, ..., vehicle n, t is the human reaction time, and k is the safety adjustment factor set according to the off-road environment.

4. The unmanned vehicle formation trajectory planning and control method according to claim 1, characterized in that: The first formation trajectory control model or the second formation trajectory control model adopts an improved model based on deep learning.

5. The unmanned vehicle formation trajectory planning and control method according to claim 4, characterized in that: The first formation trajectory control model or the second formation trajectory control model adopts improved deep learning.

6. An unmanned vehicle formation trajectory planning and control system, the system comprising a master vehicle data acquisition module, a slave vehicle status data acquisition module, a master vehicle data processing module, a formation trajectory control model building module, a formation trajectory control model updating module, and a master vehicle formation trajectory control module, characterized in that: The main vehicle data acquisition module includes a main vehicle laser radar, which is used to obtain first drivable channel information of the drivable channel between obstacles on both sides of the road, and is also used to obtain second drivable channel information of the drivable channel between obstacles on both sides of the road, and construct a coordinate system according to the main vehicle, and also obtain the first trajectory planning formation at this time; The slave vehicle status data acquisition module is wirelessly connected to the master vehicle data processing module to obtain first slave vehicle status information between the slave vehicle and the master vehicle, and also obtain second slave vehicle status information between the slave vehicle and the master vehicle; The master vehicle data processing module is configured to obtain a first safety value according to the second drivable channel information and the second slave vehicle state information; The formation trajectory control model construction module: constructs a first formation trajectory control model according to the first drivable channel information, the first slave vehicle state information and the first trajectory planning formation, and also constructs a second formation trajectory control model according to the first drivable channel information, the first slave vehicle state information combined with the second drivable channel information and the second slave vehicle state information; The formation trajectory control model updating module: obtains a first safety value according to the second drivable channel information and the second slave vehicle state information, and compares the first safety value with a set safety threshold to determine whether to add to the training of the second formation trajectory control model; The main vehicle formation trajectory control module: calls the first formation trajectory control model to process the second drivable channel information and the second slave vehicle status information to generate a second trajectory planning formation, and calls the second formation trajectory control model to process the subsequently obtained drivable channel information and slave vehicle status information to generate a real-time trajectory planning formation, and controls the driving formation of the unmanned vehicle formation according to the real-time trajectory planning formation.

7. The unmanned vehicle formation trajectory planning and control system according to claim 6, characterized in that: The first drivable channel information or the second drivable channel information includes the width d between the obstacles on both sides of the drivable channel and the angle between the host vehicle and the obstacles on both sides detected in real time. The length L and width d of the drivable channel formed by the obstacles on both sides are obtained by the distance and angle between the obstacle objects in the drivable channel and the main vehicle obtained by the laser radar of the main vehicle.

8. The unmanned vehicle formation trajectory planning and control system according to claim 7, characterized in that: The first formation trajectory control model or the second formation trajectory control model adopts an improved model based on deep learning; The first formation trajectory control model or the second formation trajectory control model adopts improved deep learning.

Images