Multi-terminal vehicle management system based on intelligent algorithm

Abstract

The invention discloses a multi-terminal vehicle management system based on an intelligent algorithm, relates to the technical field of vehicle management, and is used for solving the problem of insufficient real-time performance and cooperative efficiency of vehicle management. According to the method, a multi-modal sensor fusion technology is combined with Gaussian process regression, complex environment data is modeled in real time, dynamic target and static background features are accurately extracted, environment changes are analyzed and predicted in combination with a time sequence, accurate input is provided for path planning and task allocation, a path is dynamically adjusted through an incremental optimization algorithm, emergencies are coped, and the path planning and task allocation efficiency is improved. According to a distributed path optimization mechanism, multi-vehicle task path conflicts are eliminated, the cooperation efficiency is improved, a priority model is constructed based on the task emergency degree, the time constraint and the resource matching degree, the weight is dynamically adjusted, vehicle and task matching is optimized in combination with reinforcement learning, the task execution efficiency is ensured, resource competition and path conflicts are solved, and the task execution efficiency is improved. Efficient cooperation in a complex dynamic environment is realized, and the real-time performance and management efficiency of vehicle management are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle management, and more specifically, to a multi-terminal vehicle management system based on an intelligent algorithm. Background Art

[0002] With the rapid development of the automobile industry and the widespread application of intelligent technology, vehicle management systems are undergoing profound changes. Traditional vehicle management methods can no longer meet the high requirements of modern transportation and logistics industries for efficiency, safety and real-time performance. Multi-terminal vehicle management systems have emerged, aiming to achieve comprehensive monitoring and management of vehicles by integrating multiple intelligent terminal devices.

[0003] A multi-terminal vehicle management system usually consists of an on-board terminal, a cloud computing processing platform and a data analysis platform. The on-board terminal is responsible for collecting the real-time operation data of the vehicle, including location, speed, fuel consumption, driving behavior and other information, and transmitting the data to the cloud via a wireless network. The cloud computing processing platform analyzes massive amounts of data, generates various reports and visualization results, and provides decision support for managers.

[0004] Deficiencies in existing technologies: The vehicle management system has insufficient adaptability in complex dynamic environments, and its ability to fuse multi-source sensor data and accurately perceive dynamic environments is weak, especially in severe weather or complex terrain. It is difficult to provide reliable support, resulting in insufficient flexibility in response to emergencies in terms of path planning. Path adjustments usually require global recalculation, which is difficult to meet real-time requirements, and the path conflict problem of multi-vehicle tasks has not been effectively resolved. Vehicle scheduling mainly relies on regular allocation, and fails to comprehensively consider multi-dimensional factors such as the urgency of the task, time constraints, and vehicle status for optimization, resulting in unreasonable task allocation and failure to ensure timely completion of high-priority tasks. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, there is a solution as follows to solve the problems of insufficient real-time and collaborative efficiency of vehicle management in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: comprising a data acquisition and processing module, a dynamic path response module, and a vehicle dispatch management module, wherein the modules are connected via signals; The data acquisition and processing module is used to deploy multimodal sensors to collect environmental feature data, perform time alignment and spatial synchronization on the sensor data, extract dynamic and static features of the environment based on the sensor fusion algorithm, build a dynamic environment model based on Gaussian process regression and time series analysis, and predict environmental change trends; The dynamic path response module establishes a multi-objective optimization model to generate an initial path based on the environmental modeling results and task requirements, receives dynamic environmental change information in real time, adjusts and optimizes the initial path based on the incremental path, and performs path conflict optimization based on the distributed path optimization algorithm in multi-vehicle operation scenarios; The vehicle scheduling management module is used to quantify task priorities according to task urgency, time constraints and resource matching, screen vehicles according to vehicle status evaluation, match vehicles and tasks based on the scheduling optimization model, optimize task allocation according to conflict detection and task sequence adjustment, and optimize scheduling efficiency.

[0007] In a preferred embodiment, a multimodal sensor is deployed to collect environmental feature data, and the sensor data is time-aligned and spatially synchronized, and the dynamic and static features of the environment are extracted according to the sensor fusion algorithm. The specific steps are as follows: The sensors include laser radar, which is used to provide point cloud data to indicate the spatial position of obstacles; cameras, which are used to capture road images and identify environmental features; millimeter wave radar, which is used to detect dynamic objects around the vehicle; temperature and humidity sensors, which are used to record ambient temperature and humidity and judge special weather conditions. Each sensor independently collects features and forms a local environmental description to construct a collection of collected data. The collected sensor data are aligned and synchronized with unified timestamps, and the data are uniformly projected into the global coordinate system according to the installation location information of the sensor. The unified coordinate system is: ,in, is the position of the kth sensor data in the global coordinate system, is the rotation matrix of the kth sensor, is the original point data of the kth sensor, is the translation vector of the kth sensor; Integrate sensor data according to the sensor fusion algorithm, extract environmental features, perform point cloud and image fusion, and use the depth information of the lidar and the image of the camera for joint modeling; Separate dynamic and static features, separate moving targets and fixed backgrounds in the environment through Kalman filtering, and determine dynamic and static features.

[0008] In a preferred embodiment, a dynamic environmental model is constructed based on Gaussian process regression and time series analysis to predict environmental change trends. The specific steps include: Based on the fused environmental features, a dynamic environmental model is constructed. Based on Gaussian process regression, continuous spatial modeling of environmental data is performed, and a three-dimensional dynamic map is constructed based on time series analysis. To predict environmental changes, the environmental status prediction formula is: ,in, is the predicted environmental state of the vehicle at position s, is the kernel function between the position s and the sampling point, is the kernel matrix of the sampling points, is the noise matrix, is the observed value of the environmental state; Determine environmental change trends based on predicted environmental status.

[0009] In a preferred embodiment, based on the environmental modeling results and task requirements, a multi-objective optimization model is established to generate an initial path, dynamic environmental change information is received in real time, the initial path is optimized according to the incremental path adjustment, and in a multi-vehicle operation scenario, path conflict optimization is performed according to a distributed path optimization algorithm. The specific steps are as follows: According to the goals and constraints, the environmental modeling results and task requirements are quantified, a multi-objective optimization model is established, and the improved A* algorithm is used to generate a path that meets the basic constraints. The initial path generation formula is: ,in, is the initial path, is the path length, is the energy consumption of the path; is the environmental risk factor of the pathway; , , is the weight parameter; Receive environmental modeling results and dynamic event information, re-evaluate key nodes in the initial path, and incrementally adjust the path if path failure or optimization space is detected. In the vehicle operation scenario, share environmental data and coordinate task paths between vehicles.

[0010] In a preferred embodiment, the method for quantifying the task priority according to the task urgency, time constraint and resource matching, and selecting the vehicle according to the vehicle status evaluation comprises the following steps: Quantify task priorities based on task urgency, time constraints, and resource matching to build a task priority model and determine the task priority score; Carry out vehicle status assessment and screening, and select candidate vehicles that meet the mission requirements based on the status of the assessed vehicles.

[0011] In a preferred embodiment, the scheduling efficiency is optimized by matching vehicles and tasks based on the scheduling optimization model, adjusting and optimizing task allocation according to conflict detection and task sequence, and comprising the following steps: Allocate vehicle resources to match mission requirements based on mission priority and vehicle status, and use deep Q network learning algorithm to optimize vehicle dynamic scheduling; For path conflicts between regional vehicles, the execution order of tasks is adjusted, and tasks are executed based on conflict detection and task order adjustment.

[0012] The technical effects and advantages of the multi-terminal vehicle management system based on intelligent algorithm of the present invention are as follows: The present invention combines multimodal sensor fusion technology with Gaussian process regression to collect and model complex environmental data in real time, accurately extract dynamic target and static background features, and predict environmental change trends through time series analysis, providing accurate input for path planning and task allocation. It comprehensively considers time, energy consumption and safety constraints through a multi-objective optimization model, and uses an incremental optimization algorithm to quickly adjust the path to respond to emergencies. At the same time, a distributed path optimization mechanism is used to eliminate multi-vehicle task path conflicts and improve the efficiency of multi-terminal collaboration. A priority model is built based on task priority assessment combined with task urgency, time constraints and resource matching. The weights are dynamically adjusted to adapt to scenario requirements, and the matching strategy between vehicles and tasks is optimized based on reinforcement learning to ensure the timeliness and efficiency of task execution. Through information sharing and distributed optimization algorithms, the system coordinates multi-vehicle task allocation, solves resource competition and path conflict problems, and improves the overall task completion efficiency. It can achieve efficient collaboration in complex dynamic environments and significantly improve the real-time performance of the vehicle management system. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a structural schematic diagram of a multi-terminal vehicle management system based on an intelligent algorithm of the present invention. DETAILED DESCRIPTION

[0014] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0015] In order to achieve the above objectives, Figure 1 The structural diagram of the multi-terminal vehicle management system based on the intelligent algorithm of the present invention is given, which specifically includes a data acquisition and processing module, a dynamic path response module, and a vehicle dispatch management module, and each module is connected by a signal; The data acquisition and processing module is used to deploy multimodal sensors to collect environmental feature data, perform time alignment and spatial synchronization on the sensor data, extract dynamic and static features of the environment based on the sensor fusion algorithm, build a dynamic environment model based on Gaussian process regression and time series analysis, and predict environmental change trends; The dynamic path response module establishes a multi-objective optimization model to generate an initial path based on the environmental modeling results and task requirements, receives dynamic environmental change information in real time, adjusts and optimizes the initial path based on the incremental path, and performs path conflict optimization based on the distributed path optimization algorithm in multi-vehicle operation scenarios; The vehicle scheduling management module is used to quantify task priorities according to task urgency, time constraints and resource matching, screen vehicles according to vehicle status evaluation, match vehicles and tasks based on the scheduling optimization model, optimize task allocation according to conflict detection and task sequence adjustment, and optimize scheduling efficiency.

[0016] Step 1: Perform multimodal sensor data collection and preprocessing. The specific steps are as follows: Through multimodal perception technology, real-time collection and modeling of vehicle operating environment, deployment of multimodal sensors, each sensor has a unique collection advantage for specific environmental features. For example, LiDAR is good at measuring obstacles in three-dimensional space, cameras capture image details in the environment, millimeter-wave radar has high-precision detection capabilities for dynamic targets, and temperature and humidity sensors are used to determine the degree of icing and slippery road surface. Multimodal sensors collect environmental feature data. Specific sensors include: LiDAR, which is used to provide point cloud data to indicate the spatial position of obstacles; cameras, which are used to capture road images and identify environmental features (such as water accumulation, obstacle types); millimeter-wave radar, which is used to detect dynamic objects around the vehicle (such as pedestrians and other vehicles); temperature and humidity sensors, which are used to record ambient temperature and humidity and determine special weather (such as icing and slippery roads). Each sensor independently collects multi-dimensional features to form a local environmental description and build a collection of collected data: , collect data sets represents the n eigenvalues ​​obtained from the i-th sensor, for example, is the jth eigenvalue of the i-th sensor; Since multimodal sensors are installed at different locations, the spatial and temporal distributions of their collected data are inconsistent, so alignment and synchronization are key steps to ensure data availability; In actual scenarios, such as narrow roads in mountainous areas, vehicles need to use both the point cloud data from the lidar and the image information from the camera to determine whether there are obstacles ahead. Spatial alignment unifies the data from the lidar and camera into the vehicle's global coordinate system, and time synchronization ensures that the data from both come from the same moment. Multiple sensors have different sampling frequencies and spatial positions, so their data need to be aligned and synchronized. By unifying the timestamp, the time consistency of each sensor data is ensured, and the data is uniformly projected into the global coordinate system according to the installation location information of the sensor. The unified coordinate system is: ,in, is the position of the kth sensor data in the global coordinate system, is the rotation matrix of the kth sensor, is the original point data of the kth sensor, is the translation vector of the kth sensor; Data fusion extracts key information by integrating multimodal data. For example, on a foggy highway, the visual range of the camera is limited, but the millimeter-wave radar can penetrate the fog to provide dynamic target information, and the lidar captures the road boundary point cloud data. The fusion algorithm integrates this information into a unified set of environmental features, thereby generating a more reliable input; Integrate multimodal data through sensor fusion algorithms, extract environmental features, fuse point clouds and images, and use the depth information of lidar and the image of the camera for joint modeling; Separate dynamic and static features. Through Kalman filtering, dynamically separate the moving targets and fixed backgrounds in the environment to determine the dynamic and static features. The formula is: ,in, is the environmental feature extracted at time t after fusion, F is the fusion algorithm, which uses multimodal data to extract unified features. is the alignment data from m sensors; The fusion algorithm integrates the feature data of each sensor through a nonlinear model to eliminate the possible blind spots of single-modal sensors. Dynamic features such as pedestrians and vehicles and static features such as road boundaries and obstacle types are extracted for subsequent modeling.

[0017] Based on the fused environmental features, a dynamic environmental model is constructed to describe the global state of the environment. Based on Gaussian process regression (GPR), continuous spatial modeling of environmental data is performed, and based on time series analysis, a three-dimensional dynamic map is constructed to capture the trend of environmental changes and predict possible environmental changes. The environmental state prediction formula is: ,in, is the predicted environmental state of the vehicle at position s, is the kernel function between the position s and the sampling point, is the kernel matrix of the sampling points, is the noise matrix, is the observed value of the environmental state; By combining spatial modeling and time series analysis of Gaussian processes, environmental change trends are determined based on predicted environmental states; It should be noted that dynamic modeling needs to fully consider the correlation between time and space. Time series analysis supplements the ability to predict future environmental conditions, while the kernel function design of Gaussian process regression determines the accuracy of modeling. The kernel function needs to select an adaptive model suitable for complex terrain, such as the radial basis function (RBF) kernel, to enhance the modeling ability of nonlinear changes.

[0018] Step 2: After completing the real-time collection and dynamic modeling of the multimodal perceived environment state in step 1, the dynamic environment data and task objectives are integrated to generate the optimal path that meets the multi-dimensional constraints, and at the same time quickly respond to emergencies and adjust the path. The specific steps are as follows: Dynamic path planning needs to quantify the environmental modeling results (predicted environmental status) and task requirements (such as time constraints, energy consumption requirements, and path safety requirements) based on multiple goals (such as shortest time, minimum energy consumption, and safety) and constraints (such as road width, slope, and waterlogged area), and establish a multi-objective optimization model. In the initial stage of path planning, an improved A* algorithm is used to generate a path that meets the basic constraints. The initial path generation formula is: ,in, is the initial path, is the path length, is the energy consumption of the path; is the environmental risk factor of the pathway; , , Weight parameters are dynamically adjusted according to task priorities; Through the improved A* algorithm, the evaluation function of the node is dynamically updated when calculating the path, and the weight parameter is dynamically adjusted according to the task priority. For example, in logistics distribution, when time is the priority, weight; in bad weather, increase The weight of is used to give priority to the safe path; Receive environmental modeling results and dynamic event information (such as road congestion, sudden accidents) in real time, re-evaluate key nodes in the initial path, and dynamically adjust the path if path failure or optimization space is detected. For example, when a road is temporarily closed, the path adjustment module recalculates the pass nodes to minimize interference with task execution. Path adjustment is performed incrementally to avoid calculation delays caused by global path replanning. Multi-terminal vehicle management systems often involve multiple vehicles running at the same time. Distributed path optimization coordinates the task paths of multiple vehicles by sharing environmental data to avoid local conflicts or waste of resources. The distributed optimization algorithm formula of the path is: ,in, is the final path of the i-th vehicle, is the independent cost of the path of the i-th vehicle; is the collaboration cost between paths, indicating path conflicts or collaboration benefits. Multi-vehicle collaborative optimization reduces conflicts by sharing path information, such as reducing competition among multiple vehicles for the same road section. The collaboration cost is calculated through a distributed algorithm. Each vehicle completes path optimization at a local node while maintaining consistency with the planning results of other vehicles. Multi-objective constraint modeling and initial path planning provide the initial solution for dynamic path planning. It generates the optimal path basis by balancing multi-dimensional objectives and responds to environmental changes through incremental optimization, ensuring the flexibility and real-time performance of path planning and reducing the burden of global path recalculation. Then, it optimizes efficient collaboration under multiple terminals and multiple vehicles, improves the overall performance of the system in complex scenarios, resolves path conflicts and improves overall efficiency.

[0019] Step 3: Perform dynamic vehicle scheduling and task priority evaluation, that is, optimize the vehicle scheduling strategy according to dynamic conditions such as vehicle status, task requirements, and environmental factors to achieve efficient resource allocation and task execution. The specific steps are as follows: Task priority quantification is the basis of scheduling. Through multi-factor analysis, we build a task priority model based on the dimensions of task urgency, time constraints, resource matching, etc. The formula is: ,in, The urgency of the task; The urgency of the task time constraints; is the degree of matching of available resources, , , is the weight parameter, reflecting the importance of the task dimension; It should be noted that the matching degree of available resources indicates whether a specific resource (such as a vehicle) is suitable for performing a task, and the matching degree depends on the degree of conformity between the task requirements and the resource characteristics; It is determined by matching resource characteristics with task requirements. First, the vehicle type, such as cold chain tasks need to match refrigerated trucks, and dangerous goods transportation needs to match special vehicles. Secondly, the load capacity, such as logistics tasks need to evaluate whether the maximum load of the vehicle meets the task requirements. Then, the state characteristics, such as the vehicle's health score, power or fuel remaining, meet the task requirements. The formula is: ,in, The kth requirement dimension of the task, is the kth characteristic dimension of the vehicle, It means if and If it matches, it is 1, otherwise it is 0; The weights represent the importance of different demand dimensions; Then, the vehicle status (such as location, battery power, and mechanical performance) is dynamically sensed and the screening threshold is adjusted according to the task requirements. For example, if the task requires long-distance driving, high-endurance vehicles are given priority; Similarly, the urgency of the task time constraint indicates the strictness with which the task must be completed within the specified time. The tighter the time, the higher the weight. The urgency of the task indicates a comprehensive assessment of the task's sensitivity to time and the necessity of completion. The urgency is closely related to the task type and the actual environment.

[0020] Carry out vehicle status evaluation and availability screening. By evaluating the vehicle status, candidate vehicles that meet the mission requirements are screened. The screening formula is: ,in, For the available vehicle collection, For vehicles The distance from the mission start point, For vehicles The load capacity, For vehicles Health scores (e.g. battery charge, mechanical performance), , , For respective thresholds used for screening, the screening thresholds are dynamically set; The task priority quantitative model is dynamically adjusted in actual scenarios. For example, in a logistics scenario, the task of delivering food has a higher urgency, while ordinary goods can have a lower priority. The task priority score directly determines the sorting strategy when scheduling system resources. According to the task priority and vehicle status, vehicle resources are dynamically allocated to match the task requirements. Reinforcement learning algorithms (such as deep Q networks) are used for dynamic scheduling optimization. The scheduling decision efficiency is improved through continuous learning. The algorithm expression is as follows: ,in, is the optimal scheduling solution, is the priority score of task i, is the vehicle assigned to task i, is the objective function, which measures the overall benefit of matching tasks and vehicles; During peak hours, logistics companies need to handle a large number of orders at the same time. The system optimizes and assigns tasks to the best vehicles. For example, small packages are assigned to electric vehicles first to reduce energy consumption, while large goods are assigned to heavy trucks. In addition, the task execution history can be used as an auxiliary basis for scheduling. For example, if a vehicle has a high delivery success rate in a certain area in the past, the system tends to continue to assign tasks to that area. The deep Q network improves scheduling efficiency through feedback learning. For example, if the completion time of a task is significantly lower than the average, the system enhances the decision probability of similar task assignments through learning. The objective function not only evaluates the current scheduling benefits, but also needs to consider long-term benefits (such as balanced utilization of vehicles) to avoid excessive wear of some vehicles due to high-frequency tasks.

[0021] In multi-task scenarios, vehicle scheduling may face task conflicts or insufficient resources. For example, in logistics scenarios, multiple tasks may be concentrated in the same area, resulting in path conflicts between vehicles. By adjusting the execution order of tasks or reallocating some tasks to spare vehicles, resource waste can be reduced and scheduling efficiency can be improved. By task decomposition and scheduling priority adjustment, the execution order and resource allocation of multiple tasks are coordinated. The scheduling solution generation expression is: ,in, is a conflict-free scheduling scheme, C is the set of task conflicts, is the conflict adjustment cost for task j; For example, in logistics distribution, cold chain transportation tasks (such as frozen food) have a much higher priority than ordinary cargo distribution. By quantifying the urgency and time constraints of cold chain tasks through models, they can be given a higher priority. In addition, resource matching reflects the adaptability of tasks to specific vehicles. For example, cold chain tasks require refrigerated trucks to complete, while ordinary vehicles are not suitable for undertaking them.

[0022] Through conflict detection and task sequence adjustment, multi-task execution is coordinated, and the overall system performance is optimized through conflict adjustment, resource competition and path conflicts are avoided, and the overall delivery efficiency is optimized.

[0023] The threshold information in this embodiment is pre-set by professionals and will not be explained in detail here. Some parameter English letters in the embodiments have the same situation, but different meanings are explained when used, which will not be explained one by one here.

[0024] The present invention combines multimodal sensor fusion technology with Gaussian process regression to collect and model complex environmental data in real time, accurately extract dynamic target and static background features, and predict environmental change trends through time series analysis, providing accurate input for path planning and task allocation. It comprehensively considers time, energy consumption and safety constraints through a multi-objective optimization model, and uses an incremental optimization algorithm to quickly adjust the path to respond to emergencies. At the same time, a distributed path optimization mechanism is used to eliminate multi-vehicle task path conflicts and improve the efficiency of multi-terminal collaboration. A priority model is built based on task priority assessment combined with task urgency, time constraints and resource matching. The weights are dynamically adjusted to adapt to scenario requirements, and the matching strategy between vehicles and tasks is optimized based on reinforcement learning to ensure the timeliness and efficiency of task execution. Through information sharing and distributed optimization algorithms, the system coordinates multi-vehicle task allocation, solves resource competition and path conflict problems, and improves the overall task completion efficiency. It can achieve efficient collaboration in complex dynamic environments and significantly improve the real-time performance of the vehicle management system.

[0025] The above formulas are all dimensionless and numerical calculations. The formula is a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0026] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product.

[0027] Those of ordinary skill in the art will appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0028] In addition, each functional module in each embodiment of the present application may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0029] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0030] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-terminal vehicle management system based on intelligent algorithms, characterized by: It includes data acquisition and processing module, dynamic path response module, and vehicle dispatch management module, and each module is connected through signals; The data acquisition and processing module is used to deploy multimodal sensors to collect environmental feature data, perform time alignment and spatial synchronization on the sensor data, extract dynamic and static features of the environment based on the sensor fusion algorithm, build a dynamic environment model based on Gaussian process regression and time series analysis, and predict environmental change trends; The dynamic path response module establishes a multi-objective optimization model to generate an initial path based on the environmental modeling results and task requirements, receives dynamic environmental change information in real time, adjusts and optimizes the initial path based on the incremental path, and performs path conflict optimization based on the distributed path optimization algorithm in multi-vehicle operation scenarios; The vehicle scheduling management module is used to quantify task priorities according to task urgency, time constraints and resource matching, screen vehicles according to vehicle status evaluation, match vehicles and tasks based on the scheduling optimization model, optimize task allocation according to conflict detection and task sequence adjustment, and optimize scheduling efficiency.

2. The multi-terminal vehicle management system based on intelligent algorithm according to claim 1 is characterized by: It is used to deploy multimodal sensors to collect environmental feature data, and perform time alignment and spatial synchronization on the sensor data. The dynamic and static features of the environment are extracted according to the sensor fusion algorithm. The specific steps are as follows: The sensors include laser radar, which is used to provide point cloud data to indicate the spatial position of obstacles; cameras, which are used to capture road images and identify environmental features; millimeter wave radar, which is used to detect dynamic objects around the vehicle; temperature and humidity sensors, which are used to record ambient temperature and humidity and judge special weather conditions. Each sensor independently collects features and forms a local environmental description to construct a collection of collected data. The collected sensor data are aligned and synchronized with unified timestamps, and the data are uniformly projected into the global coordinate system according to the installation location information of the sensor. The unified coordinate system is: ,in, is the position of the kth sensor data in the global coordinate system, is the rotation matrix of the kth sensor, is the original point data of the kth sensor, is the translation vector of the kth sensor; Integrate sensor data according to the sensor fusion algorithm, extract environmental features, perform point cloud and image fusion, and use the depth information of the lidar and the image of the camera for joint modeling; Separate dynamic and static features, separate moving targets and fixed backgrounds in the environment through Kalman filtering, and determine dynamic and static features.

3. The multi-terminal vehicle management system based on intelligent algorithm according to claim 2 is characterized in that: A dynamic environmental model is constructed based on Gaussian process regression and time series analysis to predict environmental change trends. The specific steps include: Based on the fused environmental features, a dynamic environmental model is constructed. Based on Gaussian process regression, continuous spatial modeling of environmental data is performed, and a three-dimensional dynamic map is constructed based on time series analysis. To predict environmental changes, the environmental status prediction formula is: ,in, is the predicted environmental state of the vehicle at position s, is the kernel function between the position s and the sampling point, is the kernel matrix of the sampling points, is the noise matrix, is the observed value of the environmental state; Determine environmental change trends based on predicted environmental status.

4. The multi-terminal vehicle management system based on intelligent algorithm according to claim 3 is characterized by: Based on the environmental modeling results and task requirements, a multi-objective optimization model is established to generate the initial path, and dynamic environmental change information is received in real time. The initial path is optimized according to the incremental path adjustment, and path conflict optimization is performed according to the distributed path optimization algorithm in the multi-vehicle operation scenario. The specific steps are as follows: According to the objectives and constraints, the environmental modeling results and task requirements are quantified, a multi-objective optimization model is established, and the improved A* algorithm is used to generate a path that meets the basic constraints. The initial path generation formula is: ,in, is the initial path, is the path length, is the energy consumption of the path; is the environmental risk factor of the pathway; , , is the weight parameter; Receive environmental modeling results and dynamic event information, re-evaluate key nodes in the initial path, and incrementally adjust the path if path failure or optimization space is detected. In the vehicle operation scenario, share environmental data and coordinate task paths between vehicles.

5. The multi-terminal vehicle management system based on intelligent algorithm according to claim 4 is characterized by: It is used to quantify task priorities according to task urgency, time constraints and resource matching, and to select vehicles according to vehicle status evaluation, including the following steps: Quantify task priorities based on task urgency, time constraints, and resource matching to build a task priority model and determine the task priority score; Carry out vehicle status assessment and screening, and select candidate vehicles that meet the mission requirements based on the status of the assessed vehicles.

6. The multi-terminal vehicle management system based on intelligent algorithm according to claim 5 is characterized by: Match vehicles and tasks based on the scheduling optimization model, optimize task allocation according to conflict detection and task sequence adjustment, and optimize scheduling efficiency, including the following steps: Allocate vehicle resources to match mission requirements based on mission priority and vehicle status, and use deep Q network learning algorithm to optimize vehicle dynamic scheduling; For path conflicts between regional vehicles, the execution order of tasks is adjusted, and tasks are executed based on conflict detection and task order adjustment.

Images