Park networking unmanned vehicle dynamic scheduling planning method and system

Abstract

The invention relates to the technical field of unmanned vehicle scheduling, in particular to a park networking unmanned vehicle dynamic scheduling planning method and system, and the method comprises the steps: collecting dynamic obstacle data through a multi-source sensor and V2X equipment, generating a four-dimensional trafficability map, formulating an avoidance strategy weight, optimizing a multi-vehicle track, and carrying out the optimization of a multi-vehicle track. And evaluating a threat value to trigger the reconfiguration of the resident area by combining the edge node and the cloud adjusting path. The system comprises a sensing module, a decision-making module, a storage module, an allocation module, a regulation module and a verification module, and an exception handling module is linked with a threat value change rate to optimize resource allocation. According to the method, the avoidance strategy can be dynamically adjusted according to the threat value, the scheduling accuracy and the resource allocation accuracy are improved, the vehicle safety is guaranteed, and the task failure rate is reduced.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation and autonomous driving technologies, and in particular to a method and system for dynamic scheduling planning of networked unmanned vehicles in a campus. Background Art

[0002] Currently, networked unmanned vehicle dispatching systems in campuses generally rely on preset path planning and static parking zone allocation strategies. While this improves vehicle operating efficiency to a certain extent, it can easily lead to path planning instability due to the combined interference of dynamic obstacles and conflicting trajectories of multiple vehicles. To improve the stability and adaptability of the dispatching system, corresponding measures are needed to address issues such as sudden obstacles intruding into the preset parking zone, insufficient capacity caused by multiple vehicles simultaneously applying for parking, and conflicts between the parking zone and the movement trajectory of dynamic obstacles.

[0003] Therefore, a dynamic scheduling method based on a spatiotemporal grid map can be used to optimize the operation of autonomous vehicles. When an unexpected obstacle intrudes into a parking zone, the system maintains system stability by adjusting the route and parking zone allocation in real time. This is typically handled through a distributed decision-making unit. However, unexpected obstacles intruding into a pre-set parking zone can trigger frequent path replanning. Simultaneous parking requests by multiple vehicles can also lead to increased competition for parking zone resources, impacting overall traffic efficiency. Conflicts between the parking zone and dynamic obstacle trajectories can cause localized congestion or even mission failure.

[0004] However, conventional dynamic scheduling methods are less applicable in complex scenarios. Campus environments, due to the random nature of dynamic obstacles, the need for multi-vehicle coordination, and spatial limitations, place higher demands on scheduling systems. For example, when sudden construction vehicles occupy spaces or when a large number of vehicles apply to park during peak hours, existing scheduling methods are prone to uneven resource allocation, resulting in long wait times for some vehicles. Furthermore, conflict resolution strategies are inefficient, making them more susceptible to system oscillation.

[0005] For example, the combined interference of dynamic obstacles and conflicting trajectories of multiple vehicles leads to unstable path planning. Problems such as sudden obstacles intruding into pre-set parking areas, insufficient capacity due to multiple vehicles simultaneously requesting parking, and conflicts between parking areas and the motion trajectories of dynamic obstacles further highlight the limitations of existing technologies. These situations indicate that existing technologies need to be further optimized to handle complex dynamic scenarios, in order to enhance the robustness and adaptability of dispatch systems. Summary of the Invention

[0006] The purpose of the present invention is to overcome the deficiencies in the prior art, solve or at least alleviate the problem of path planning instability of networked unmanned vehicles in a campus under the combined interference of dynamic obstacles and conflicts between multiple vehicle trajectories, and provide a dynamic scheduling planning method and system for networked unmanned vehicles in a campus.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for dynamic scheduling planning of networked unmanned vehicles in a park, comprising the following steps: S1. Collect 3D trajectory data of dynamic obstacles through multi-source sensors and V2X communication equipment, and generate a 4D accessibility map based on the spatiotemporal grid map. S2. Develop different avoidance strategy weights based on the dynamic obstacle's motion state, the unmanned vehicle's planned arrival time, remaining battery power, and the importance of the cargo it carries. S3, applying the obtained avoidance strategy weights to the distributed decision-making unit to optimize the multi-vehicle trajectories through a mixed integer programming model and asymmetric game strategy; S4, the edge node monitors the vehicle trajectory execution in real time and adjusts the path based on the global optimization results in the cloud; S5. While determining the avoidance strategy weight, the dynamic threat value of the resident area is evaluated. When the threat value exceeds the set threshold, the resident area reconstruction process is triggered.

[0008] In order to further realize the present invention, the following technical solutions may be preferably used: Preferably, in step S5, the comprehensive threat value is calculated by the movement direction and speed of the dynamic obstacle and the distance between the dynamic obstacle and the preset residence area; When the threat value is less than the first determination value, it is determined as a low threat area; when the threat value is between the first determination value and the second determination value, it is determined as a medium threat area; when the threat value is greater than the second determination value, it is determined as a high threat area.

[0009] Preferably, the step S5 further includes the following steps: When the threat value is in the low threat area and the length of the residence application queue is less than the first judgment value, the avoidance strategy weight is changed; When the threat value is in a high threat area and the length of the residence application queue is greater than the second judgment value, the avoidance strategy weight is changed; Change the avoidance strategy weight when the threat value is in the medium threat area and the length of the residence application queue is stable; The avoidance strategy weight when the threat value is in the low threat area is greater than the avoidance strategy weight when the threat value is in the high threat area, and the avoidance strategy weight when the threat value is in the high threat area is greater than the avoidance strategy weight when the threat value is in the medium threat area.

[0010] Preferably, in step S3, the avoidance strategy includes an emergency detour weight, a deceleration and residence weight, and a lateral offset weight, and the emergency detour weight, the deceleration and residence weight, and the lateral offset weight are multiplied by the avoidance strategy weight respectively to obtain an actual avoidance strategy weight.

[0011] Preferably, the step S3 further includes the following steps: Compare the real-time position of the dynamic obstacle with the emergency detour weight and the deceleration and residence weight. When the real-time position threat value is greater than the deceleration and residence weight, enter the lateral deviation state. When the real-time location threat value is less than the emergency detour weight, the planned arrival time of the unmanned vehicle is compared with the deceleration and residence weight. When the planned arrival time is less than the deceleration and residence weight, the vehicle enters the deceleration and residence state. When the planned arrival time is greater than the deceleration and residence weight, the vehicle enters the emergency detour state. When the real-time position threat value is greater than the emergency detour weight and less than the deceleration and residence weight, the planned arrival time of the unmanned vehicle is compared with the deceleration and residence weight. When the planned arrival time is less than the deceleration and residence weight, it enters the lateral deviation state. When the planned arrival time is greater than the deceleration and residence weight, it enters the deceleration and residence state.

[0012] Preferably, the step S5 further includes the following steps: The real-time position of dynamic obstacles is obtained through V2X communication equipment, and the threat value change rate of the current time window is obtained. The threat value change rate of each time window is recorded in sequence. Compare the threat value change rate of the current time window with the threat value change rates of the previous multiple time windows. When the threat value change rates of multiple time windows are not all positive and do not all exceed the set value, maintain the resident zone allocation state; When the threat value change rates of multiple time windows are all positive and exceed the set value, the resident area allocation state is changed to the emergency reconstruction state, and the resident area allocation state remains in the emergency reconstruction state in the first subsequent time window; The state of the resident zone allocation in the subsequent second time window is changed according to the rate of change of the threat value in the subsequent first time window and the second time window. Preferably, the step S5 further includes the following steps: Determine the positive or negative state of the threat value change rate in the first subsequent time window. If the threat value change rate is positive, determine the positive or negative state of the threat value change rate in the second subsequent time window. If the threat value change rate is positive, the resident area allocation state remains in the emergency reconstruction state. If the threat value change rate is negative, determine the absolute value of the threat value change rate. If the absolute value of the threat value change rate is less than the set value, maintain the emergency reconstruction state. If the absolute value of the threat value change rate is greater than the set value, change to the original resident area allocation state. If the threat value change rate is negative, the positive or negative state of the threat value change rate in the subsequent second time window is determined. If the threat value change rate is negative, the resident area allocation state is changed to the original resident area allocation state. If the threat value change rate is positive, the absolute value of the threat value change rate is determined. If the absolute value of the threat value change rate is greater than the set value, the emergency reconstruction state is maintained. If the absolute value of the threat value change rate is less than the set value, the state is changed to the original resident area allocation state.

[0013] Preferably, the following steps are also performed while evaluating the threat value change rate in step S5: The intrusion depth of dynamic obstacles within the monitoring time window is changed to the emergency reconstruction state when the intrusion depth exceeds the warning threshold.

[0014] A campus networked unmanned vehicle dynamic scheduling planning system, used for the above-mentioned dynamic scheduling planning method, includes a perception module, a decision module, a storage module, an allocation module, a control module and a verification module; The perception module integrates lidar, millimeter-wave radar, and cameras, and constructs a four-dimensional motion model of dynamic obstacles through timestamp alignment; The decision-making module is deployed on edge nodes, and multi-vehicle trajectory collaborative optimization is achieved through trajectory optimization computing clusters; The storage module stores three types of avoidance strategies and spatiotemporal grid maps, and maintains data consistency through a dynamic update mechanism; The allocation module is deployed on edge nodes and implements dynamic buffer management through the resident zone allocation engine; The control module realizes the dynamic coupling of manual configuration and adaptive optimization, and optimizes the resident area allocation through the weight adjustment algorithm; The verification module supports SUMO+ROS joint simulation testing and verifies trajectory stability through digital twin technology.

[0015] Preferably, it further comprises an exception handling module, which is connected in parallel to the allocation module; In the emergency reconstruction state, the response frequency of the exception handling module is linked to the threat value change rate to suppress local congestion and accelerate the release of resources in the resident area.

[0016] The beneficial effects of the present invention are: The present invention uses dynamic obstacle threat value recognition technology to distinguish park operation scenarios, dynamically adjusts the actual avoidance strategy weight according to the threat value type, and enters the scheduling state of different scenarios according to the actual avoidance strategy weight, thereby improving the accuracy of scheduling actions, reducing the misjudgment rate while ensuring vehicle safety, and reducing the mission failure rate.

[0017] At the same time, the present invention evaluates the threat value change rate to correct the resident area allocation status, predicts the intrusion of dynamic obstacles, adjusts the scheduling status based on the characteristics of the park environment, and ensures vehicle safety.

[0018] In addition, the dynamic scheduling planning system of the present invention makes matching adjustments according to different scheduling states, which not only plays a better scheduling role, but also improves the accuracy of resource allocation. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1This is a flow chart of the dynamic scheduling planning method of the present invention.

[0020] Figure 2 This is a flow chart of step S5 of the present invention.

[0021] Figure 3 This is a flowchart of the avoidance strategy weight adjustment of the present invention.

[0022] Figure 4 This is a flowchart of the threat value change rate assessment of the present invention.

[0023] Figure 5 This is a block diagram of the avoidance strategy weight calculation of the present invention. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0025] Example 1 Dispatching systems for connected autonomous vehicles in campuses generally rely on preset path planning and static parking zone allocation strategies. While this improves vehicle efficiency to a certain extent, it can easily lead to path planning instability due to the combined interference of dynamic obstacles and conflicting trajectories of multiple vehicles. To improve the stability and adaptability of the dispatching system, appropriate measures are needed to address issues such as sudden obstacles intruding into preset parking zones, insufficient parking capacity caused by multiple vehicles simultaneously applying for parking, and conflicts between parking zones and the movement trajectories of dynamic obstacles.

[0026] Conventional dynamic scheduling methods are less applicable in complex scenarios. Campus environments, due to the random nature of dynamic obstacles, the need for multi-vehicle coordination, and spatial limitations, place higher demands on scheduling systems. For example, when sudden construction vehicles occupy spaces or when a large number of vehicles apply for parking during peak hours, existing scheduling methods are prone to uneven resource allocation, resulting in long wait times for some vehicles. Furthermore, conflict resolution strategies are inefficient, making them more susceptible to system oscillation.

[0027] This embodiment discloses an invention that provides a dynamic scheduling and planning system for networked unmanned vehicles in a park. In actual applications, the system operates on unmanned vehicles in the park and realizes dynamic scheduling and path optimization through the collaborative work of multiple modules.

[0028] A dynamic scheduling and planning system for networked unmanned vehicles in a park, including a perception module, a decision module, a storage module, a distribution module, a control module, and a verification module; The perception module integrates lidar, millimeter-wave radar, and cameras, and constructs a four-dimensional motion model of dynamic obstacles through timestamp alignment; The decision-making module is deployed on edge nodes, and multi-vehicle trajectory collaborative optimization is achieved through trajectory optimization computing clusters; The storage module stores three types of avoidance strategies and spatiotemporal grid maps, and maintains data consistency through a dynamic update mechanism; The allocation module is deployed on edge nodes and implements dynamic buffer management through the resident zone allocation engine; The control module realizes the dynamic coupling of manual configuration and adaptive optimization, and optimizes the resident area allocation through the weight adjustment algorithm; The verification module supports SUMO+ROS joint simulation testing and verifies trajectory stability through digital twin technology.

[0029] It also includes an exception handling module, which is connected in parallel to the distribution module; In the emergency reconstruction state, the response frequency of the exception handling module is linked to the threat value change rate to suppress local congestion and accelerate the release of resources in the resident area.

[0030] In a real-world application scenario, consider an unmanned vehicle carrying goods within a park and encountering multiple dynamic obstacles. The perception module collects 3D trajectory data of the obstacles using multi-source sensors and uploads this data to the decision module. The decision module calculates the optimal avoidance strategy weight based on the obstacle's motion state and the unmanned vehicle's mission priority, and passes the result to the allocation module. The allocation module dynamically adjusts the allocation of resident zones based on the avoidance strategy data provided by the storage module. Simultaneously, the control module monitors resident zone usage in real time and adjusts resident zone allocation based on the rate of change in threat values. The verification module simulates and verifies the entire process using digital twin technology to ensure the stability and security of path planning. Throughout this process, the exception handling module remains on standby. If it detects an abnormal rate of change in threat values or an intrusion depth exceeding the specified limit, it immediately initiates an emergency reconfiguration process to ensure the safe operation of the unmanned vehicle.

[0031] Example 2 Reference Figure 1 - Figure 5 In order to make the above system work better, this embodiment provides a method for dynamic scheduling planning of networked unmanned vehicles in a campus, which includes the following steps: S1. Collect 3D trajectory data of dynamic obstacles through multi-source sensors and V2X communication equipment, and generate a 4D accessibility map based on the spatiotemporal grid map. S2. Develop different avoidance strategy weights based on the dynamic obstacle's motion state, the unmanned vehicle's planned arrival time, remaining battery power, and the importance of the cargo it carries. S3, applying the obtained avoidance strategy weights to the distributed decision-making unit to optimize the multi-vehicle trajectories through a mixed integer programming model and asymmetric game strategy; S4, the edge node monitors the vehicle trajectory execution in real time and adjusts the path based on the global optimization results in the cloud; S5. While determining the avoidance strategy weight, the dynamic threat value of the resident area is evaluated. When the threat value exceeds the set threshold, the resident area reconstruction process is triggered.

[0032] In step S5, the comprehensive threat value is calculated based on the movement direction and speed of the dynamic obstacle and its distance from the preset residence area; When the threat value is less than the first determination value, it is determined as a low threat area; when the threat value is between the first determination value and the second determination value, it is determined as a medium threat area; when the threat value is greater than the second determination value, it is determined as a high threat area.

[0033] The step S5 further comprises the following steps: When the threat value is in the low threat area and the length of the residence application queue is less than the first judgment value, the avoidance strategy weight is changed; When the threat value is in a high threat area and the length of the residence application queue is greater than the second judgment value, the avoidance strategy weight is changed; Change the avoidance strategy weight when the threat value is in the medium threat area and the length of the residence application queue is stable; The avoidance strategy weight when the threat value is in the low threat area is greater than the avoidance strategy weight when the threat value is in the high threat area, and the avoidance strategy weight when the threat value is in the high threat area is greater than the avoidance strategy weight when the threat value is in the medium threat area.

[0034] In step S3, the avoidance strategy includes an emergency detour weight, a deceleration and residence weight, and a lateral offset weight. The emergency detour weight, the deceleration and residence weight, and the lateral offset weight are multiplied by the avoidance strategy weight to obtain an actual avoidance strategy weight.

[0035] The step S3 further comprises the following steps: Compare the real-time position of the dynamic obstacle with the emergency detour weight and the deceleration and residence weight. When the real-time position threat value is greater than the deceleration and residence weight, enter the lateral deviation state. When the real-time location threat value is less than the emergency detour weight, the planned arrival time of the unmanned vehicle is compared with the deceleration and residence weight. When the planned arrival time is less than the deceleration and residence weight, the vehicle enters the deceleration and residence state. When the planned arrival time is greater than the deceleration and residence weight, the vehicle enters the emergency detour state. When the real-time position threat value is greater than the emergency detour weight and less than the deceleration and residence weight, the planned arrival time of the unmanned vehicle is compared with the deceleration and residence weight. When the planned arrival time is less than the deceleration and residence weight, it enters the lateral deviation state. When the planned arrival time is greater than the deceleration and residence weight, it enters the deceleration and residence state.

[0036] The step S5 further comprises the following steps: The real-time position of dynamic obstacles is obtained through V2X communication equipment, and the threat value change rate of the current time window is obtained. The threat value change rate of each time window is recorded in sequence. Compare the threat value change rate of the current time window with the threat value change rates of the previous multiple time windows. When the threat value change rates of multiple time windows are not all positive and do not all exceed the set value, maintain the resident zone allocation state; When the threat value change rates of multiple time windows are all positive and exceed the set value, the resident area allocation state is changed to the emergency reconstruction state, and the resident area allocation state remains in the emergency reconstruction state in the first subsequent time window; The state of the resident zone allocation in the subsequent second time window is changed according to the rate of change of the threat value in the subsequent first time window and the second time window. The step S5 further comprises the following steps: Determine the positive or negative state of the threat value change rate in the first subsequent time window. If the threat value change rate is positive, determine the positive or negative state of the threat value change rate in the second subsequent time window. If the threat value change rate is positive, the resident area allocation state remains in the emergency reconstruction state. If the threat value change rate is negative, determine the absolute value of the threat value change rate. If the absolute value of the threat value change rate is less than the set value, maintain the emergency reconstruction state. If the absolute value of the threat value change rate is greater than the set value, change to the original resident area allocation state. If the threat value change rate is negative, the positive or negative state of the threat value change rate in the subsequent second time window is determined. If the threat value change rate is negative, the resident area allocation state is changed to the original resident area allocation state. If the threat value change rate is positive, the absolute value of the threat value change rate is determined. If the absolute value of the threat value change rate is greater than the set value, the emergency reconstruction state is maintained. If the absolute value of the threat value change rate is less than the set value, the state is changed to the original resident area allocation state.

[0037] In step S5, while evaluating the threat value change rate, the following steps are also performed: The intrusion depth of dynamic obstacles within the monitoring time window is changed to the emergency reconstruction state when the intrusion depth exceeds the warning threshold.

[0038] The present invention uses dynamic obstacle threat value recognition technology to distinguish park operation scenarios, dynamically adjusts the actual avoidance strategy weight according to the threat value type, and enters the scheduling state of different scenarios according to the actual avoidance strategy weight, thereby improving the accuracy of scheduling actions, reducing the misjudgment rate while ensuring vehicle safety, and reducing the mission failure rate.

[0039] At the same time, the present invention evaluates the threat value change rate to correct the resident area allocation status, predicts the intrusion of dynamic obstacles, adjusts the scheduling status based on the characteristics of the park environment, and ensures vehicle safety.

[0040] In addition, the dynamic scheduling planning system of the present invention makes matching adjustments according to different scheduling states, which not only plays a better scheduling role, but also improves the accuracy of resource allocation.

[0041] Example 3 In order to better enable relevant personnel in this technical field to fully understand and implement the present invention, the specific implementation principle of the present invention is further supplemented below in combination with a specific application scenario.

[0042] During the operation of the campus's networked unmanned vehicle dynamic dispatch system, the perception module first uses lidar, millimeter-wave radar, and cameras to collaboratively collect real-time location information on dynamic obstacles within the campus. The lidar generates high-precision three-dimensional point cloud data, the millimeter-wave radar detects the speed and distance of obstacles, and the camera identifies and classifies the appearance of obstacles. The data from these sensors is timestamped and uploaded to the edge node, which generates a four-dimensional traversability map based on the spatiotemporal grid map. This map not only contains the spatial location information of dynamic obstacles, but also covers their speed, direction, and temporal trends, providing basic data support for subsequent decision-making modules. The perception module serves as the core input component of the entire system, and its output is directly transmitted to the decision-making module.

[0043] After entering the decision module, the system calculates the optimal avoidance strategy weight based on the dynamic obstacle's motion state and the autonomous vehicle's mission priority. For example, if the dynamic obstacle's combined threat value exceeds the deceleration and retention weight, the system shifts the autonomous vehicle to a lateral offset state; if the threat value is less than the emergency detour weight and the planned arrival time is less than the deceleration and retention weight, the system shifts the autonomous vehicle to a deceleration and retention state; and if the planned arrival time is greater than the deceleration and retention weight, the system shifts the autonomous vehicle to an emergency detour state. The decision module optimizes multi-vehicle trajectory coordination through a mixed integer programming model and an asymmetric game strategy, ensuring efficient vehicle operation in complex scenarios.

[0044] The storage module's primary function is to store avoidance strategies and spatiotemporal grid maps, maintaining data consistency through a dynamic update mechanism. For example, if the threat value in a particular area continues to rise, the storage module automatically adjusts the avoidance strategy weight for that area and synchronizes the updated data with the allocation module. The allocation module dynamically adjusts the allocation of resident zones based on the output of the decision module and the avoidance strategy data provided by the storage module. For example, if the threat value is in the low threat zone and the length of the resident application queue is less than a first threshold, the allocation module adjusts the avoidance strategy weight. If the threat value is in the high threat zone and the length of the resident application queue is greater than a second threshold, the allocation module also adjusts the avoidance strategy weight. Furthermore, if the threat value is in the medium threat zone and the length of the resident application queue is stable, the allocation module also adjusts the avoidance strategy weight accordingly. This dynamic adjustment mechanism ensures efficient utilization of resident zone resources.

[0045] The control module's function is to dynamically couple manual configuration with adaptive optimization, optimizing resident zone allocation through a weight adjustment algorithm. For example, when the threat value change rate across multiple time windows is positive and exceeds a set value, the control module changes the resident zone allocation state to an emergency reconstruction state, maintaining this state for the first subsequent time window. Furthermore, the control module supports linked responses from the exception handling module. When the threat value change rate is negative and its absolute value is less than a set value, the control module changes the resident zone allocation state back to the original state. The control module's user interface is simple and intuitive, making manual intervention easy for operators.

[0046] The verification module uses SUMO and ROS co-simulation testing, combined with digital twin technology, to verify trajectory stability. The verification module is connected to the decision-making and control modules via a dedicated communication protocol, providing real-time access to the UAV's trajectory execution and resident zone allocation status. By analyzing simulation results, the verification module evaluates the accuracy and stability of path planning and provides feedback to the decision-making and control modules for further optimization of scheduling strategies. For example, in a specific scenario, if the verification module discovers that the UAV's lateral excursion frequency is too high, potentially affecting mission completion efficiency, the verification module will send adjustment recommendations to the decision-making module to reduce the frequency of lateral excursion weighting.

[0047] Throughout the entire process, the exception handling module remains on standby. If it detects that the intrusion depth of a dynamic obstacle within a certain time window exceeds the warning threshold, it immediately triggers a change in the parking zone allocation status and notifies the control module to make appropriate adjustments. For example, if the intrusion depth of a dynamic obstacle exceeds the warning threshold, the exception handling module will respond quickly, reducing local congestion and accelerating the release of parking zone resources, thereby ensuring vehicle safety.

[0048] In a real-world application scenario, consider an unmanned vehicle carrying goods within a park and encountering multiple dynamic obstacles. The perception module collects 3D trajectory data of the obstacles using multi-source sensors and uploads this data to the decision module. The decision module calculates the optimal avoidance strategy weight based on the obstacle's motion state and the unmanned vehicle's mission priority, and passes the result to the allocation module. The allocation module dynamically adjusts the allocation of resident zones based on the avoidance strategy data provided by the storage module. Simultaneously, the control module monitors resident zone usage in real time and adjusts resident zone allocation based on the rate of change in threat values. The verification module simulates and verifies the entire process using digital twin technology to ensure the stability and security of path planning. Throughout this process, the exception handling module remains on standby. If it detects an abnormal rate of change in threat values or an intrusion depth exceeding the specified limit, it immediately initiates an emergency reconfiguration process to ensure the safe operation of the unmanned vehicle.

[0049] The above are only preferred specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solutions and inventive concepts of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A method for dynamic scheduling and planning of networked unmanned vehicles in a park, characterized by: The following steps are involved: S1. Collect 3D trajectory data of dynamic obstacles through multi-source sensors and V2X communication equipment, and generate a 4D accessibility map based on the spatiotemporal grid map. S2. Develop different avoidance strategy weights based on the dynamic obstacle's motion state, the unmanned vehicle's planned arrival time, remaining battery power, and the importance of the cargo it carries. S3, applying the obtained avoidance strategy weights to the distributed decision-making unit to optimize the multi-vehicle trajectories through a mixed integer programming model and asymmetric game strategy; S4, the edge node monitors the vehicle trajectory execution in real time and adjusts the path based on the global optimization results in the cloud; S5. While determining the avoidance strategy weight, the dynamic threat value of the resident area is evaluated. When the threat value exceeds the set threshold, the resident area reconstruction process is triggered.

2. The method for dynamic scheduling and planning of networked unmanned vehicles in a park according to claim 1, characterized in that: In step S5, the comprehensive threat value is calculated based on the movement direction and speed of the dynamic obstacle and its distance from the preset residence area; When the threat value is less than the first determination value, it is determined as a low threat area; when the threat value is between the first determination value and the second determination value, it is determined as a medium threat area; when the threat value is greater than the second determination value, it is determined as a high threat area.

3. The method for dynamic scheduling and planning of networked unmanned vehicles in a park according to claim 2, characterized in that: The step S5 further comprises the following steps: When the threat value is in the low threat area and the length of the residence application queue is less than the first judgment value, the avoidance strategy weight is changed; When the threat value is in a high threat area and the length of the residence application queue is greater than the second judgment value, the avoidance strategy weight is changed; Change the avoidance strategy weight when the threat value is in the medium threat area and the length of the residence application queue is stable; The avoidance strategy weight when the threat value is in the low threat area is greater than the avoidance strategy weight when the threat value is in the high threat area, and the avoidance strategy weight when the threat value is in the high threat area is greater than the avoidance strategy weight when the threat value is in the medium threat area.

4. The method for dynamic scheduling and planning of networked unmanned vehicles in a park according to claim 1, characterized in that: In step S3, the avoidance strategy includes an emergency detour weight, a deceleration and residence weight, and a lateral offset weight. The emergency detour weight, the deceleration and residence weight, and the lateral offset weight are multiplied by the avoidance strategy weight to obtain an actual avoidance strategy weight.

5. The method for dynamic scheduling and planning of networked unmanned vehicles in a park according to claim 4, characterized in that: The step S3 further comprises the following steps: Compare the real-time position of the dynamic obstacle with the emergency detour weight and the deceleration and residence weight. When the real-time position threat value is greater than the deceleration and residence weight, enter the lateral deviation state. When the real-time location threat value is less than the emergency detour weight, the planned arrival time of the unmanned vehicle is compared with the deceleration and residence weight. When the planned arrival time is less than the deceleration and residence weight, the vehicle enters the deceleration and residence state. When the planned arrival time is greater than the deceleration and residence weight, the vehicle enters the emergency detour state. When the real-time position threat value is greater than the emergency detour weight and less than the deceleration and residence weight, the planned arrival time of the unmanned vehicle is compared with the deceleration and residence weight. When the planned arrival time is less than the deceleration and residence weight, it enters the lateral deviation state. When the planned arrival time is greater than the deceleration and residence weight, it enters the deceleration and residence state.

6. The method for dynamic scheduling and planning of networked unmanned vehicles in a park according to claim 1, characterized in that: The step S5 further comprises the following steps: The real-time position of dynamic obstacles is obtained through V2X communication equipment, and the threat value change rate of the current time window is obtained. The threat value change rate of each time window is recorded in sequence. Compare the threat value change rate of the current time window with the threat value change rates of the previous multiple time windows. When the threat value change rates of multiple time windows are not all positive and do not all exceed the set value, maintain the resident zone allocation state; When the threat value change rates of multiple time windows are all positive and exceed the set value, the resident area allocation state is changed to the emergency reconstruction state, and the resident area allocation state remains in the emergency reconstruction state in the first subsequent time window; The state of the resident zone allocation in the subsequent second time window is changed according to the rate of change of the threat value in the subsequent first time window and the second time window.

7. The method for dynamic scheduling and planning of networked unmanned vehicles in a park according to claim 6, characterized in that: The step S5 further comprises the following steps: Determine the positive or negative state of the threat value change rate in the first subsequent time window. If the threat value change rate is positive, determine the positive or negative state of the threat value change rate in the second subsequent time window. If the threat value change rate is positive, the resident area allocation state remains in the emergency reconstruction state. If the threat value change rate is negative, determine the absolute value of the threat value change rate. If the absolute value of the threat value change rate is less than the set value, maintain the emergency reconstruction state. If the absolute value of the threat value change rate is greater than the set value, change to the original resident area allocation state. If the threat value change rate is negative, the positive or negative state of the threat value change rate in the subsequent second time window is determined. If the threat value change rate is negative, the resident area allocation state is changed to the original resident area allocation state. If the threat value change rate is positive, the absolute value of the threat value change rate is determined. If the absolute value of the threat value change rate is greater than the set value, the emergency reconstruction state is maintained. If the absolute value of the threat value change rate is less than the set value, the state is changed to the original resident area allocation state.

8. The method for dynamic scheduling and planning of networked unmanned vehicles in a park according to claim 6, characterized in that: In step S5, while evaluating the threat value change rate, the following steps are also performed: The intrusion depth of dynamic obstacles within the monitoring time window is changed to the emergency reconstruction state when the intrusion depth exceeds the warning threshold.

9. A dynamic scheduling and planning system for networked unmanned vehicles in a park, used in the dynamic scheduling and planning method for networked unmanned vehicles in a park as described in any one of claims 1 to 8, characterized in that: It includes perception module, decision module, storage module, allocation module, control module and verification module; The perception module integrates lidar, millimeter-wave radar, and cameras, and constructs a four-dimensional motion model of dynamic obstacles through timestamp alignment; The decision-making module is deployed on edge nodes, and multi-vehicle trajectory collaborative optimization is achieved through trajectory optimization computing clusters; The storage module stores three types of avoidance strategies and spatiotemporal grid maps, and maintains data consistency through a dynamic update mechanism; The allocation module is deployed on edge nodes and implements dynamic buffer management through the resident zone allocation engine; The control module realizes the dynamic coupling of manual configuration and adaptive optimization, and optimizes the resident area allocation through the weight adjustment algorithm; The verification module supports SUMO+ROS joint simulation testing and verifies trajectory stability through digital twin technology.

10. The campus networked unmanned vehicle dynamic scheduling planning system according to claim 9 is characterized in that: It also includes an exception handling module, which is connected in parallel to the distribution module; In the emergency reconstruction state, the response frequency of the exception handling module is linked to the threat value change rate to suppress local congestion and accelerate the release of resources in the resident area.

Images