A motion control method for unmanned vehicles based on regional visual relative positioning

Abstract

The present invention discloses an unmanned vehicle motion control method based on regional visual relative positioning, which belongs to the field of automatic driving technology, including: a drone performs coordinated motion with an unmanned vehicle based on a preset following mode, shoots the driving environment of the unmanned vehicle based on a preset time interval, and generates a first image set; performs image preprocessing on any first image in the first image set, extracts image features, and determines the current position and motion posture of the unmanned vehicle based on the image features; a control unit establishes a motion map model based on the current position and motion posture, the motion map model calculates the optimal motion path, and sends a motion instruction to the unmanned vehicle based on the optimal motion path; the unmanned vehicle moves after receiving the motion instruction, monitors the actual motion path of the unmanned vehicle in real time, dynamically adjusts the motion instruction, and completes the motion control process of the unmanned vehicle. The present invention can improve the timeliness, safety and accuracy of unmanned vehicle control.

Description

Technical Field

[0001] The present invention belongs to the technical field of autonomous driving, and in particular relates to a motion control method for an unmanned vehicle based on regional visual relative positioning. Background Art

[0002] With the rapid development of automation and intelligent technology, unmanned vehicles (UV) have become a research hotspot in the field of modern science and technology. Unmanned vehicles have shown broad application prospects in many fields such as military reconnaissance, cargo transportation, environmental monitoring, disaster relief, and autonomous driving. In order to achieve autonomous navigation and precise control of unmanned vehicles, their positioning accuracy and motion control strategy are crucial.

[0003] Traditional unmanned vehicle positioning technology mainly relies on the Global Positioning System (GPS) to determine the location of the vehicle. However, in certain specific environments, such as urban canyons, indoor spaces, tunnels or remote areas, GPS signals may be blocked or interfered with, resulting in reduced positioning accuracy or complete failure. In addition, GPS provides absolute positioning information, while in complex environments, relative positioning information is equally important for unmanned vehicle path planning and obstacle avoidance.

[0004] In recent years, with the advancement of computer vision technology, vision-based positioning methods have gradually become a hot topic of research. Visual positioning systems can estimate the position and posture changes of unmanned vehicles by analyzing the image sequences captured by the camera, extracting feature points and tracking their movements. Visual positioning systems have the advantages of low cost and rich information, but their accuracy and robustness are affected by environmental texture, lighting conditions and dynamic targets.

[0005] In actual situations, it is necessary to use image processing and computer vision technology to achieve regional visual relative positioning and perform motion control on unmanned vehicles in environments where GPS signals are limited or unavailable. Summary of the invention

[0006] In order to solve the above problems, the present invention provides an unmanned vehicle motion control method based on regional visual relative positioning to solve the problems in the prior art.

[0007] In order to achieve the above-mentioned invention object, the present invention proposes an unmanned vehicle motion control method based on regional visual relative positioning, comprising:

[0008] S1: The drone moves in coordination with the unmanned vehicle based on a preset following mode, and the camera carried by the drone photographs the driving environment of the unmanned vehicle based on a preset time interval to generate a first image set;

[0009] S2: performing image preprocessing on any first image in the first image set, extracting image features, and determining the current position and motion posture of the unmanned vehicle based on the image features;

[0010] S3: The control unit establishes a motion map model based on the current position and the motion posture, the motion map model outputs an optimal motion path, sends a motion instruction to the unmanned vehicle based on the optimal motion path, and the unmanned vehicle moves after receiving the motion instruction;

[0011] S4: monitor the actual motion path of the unmanned vehicle in real time, and dynamically adjust the motion instruction based on the actual motion path and the optimal motion path to complete the motion control process of the unmanned vehicle.

[0012] Furthermore, the determining the current position and motion posture of the unmanned vehicle based on the image features comprises the following steps:

[0013] After acquiring the regional image, the drone descends to a preset navigation point to photograph the driving environment, wherein the first image set includes first images corresponding to the front, rear, left, and right directions of the unmanned vehicle respectively;

[0014] Performing feature extraction on the first image after image enhancement to obtain the image features, matching the image features with the preset navigation points to generate matching features, and identifying the relative position of the unmanned vehicle based on the matching features;

[0015] Calculating the actual position of the unmanned vehicle based on the flight data of the unmanned aerial vehicle, the matching features and the relative position, and setting the actual position as the current position;

[0016] The movement vector of any feature point group in the matching feature is obtained based on the preset time interval, all the movement vectors are fitted to generate a target vector, and the vector direction of the target vector is set as the motion posture.

[0017] Furthermore, based on the area image, any two adjacent preset navigation points are connected to generate multiple connecting lines, the minimum angle between any two adjacent connecting lines is calculated, and the driving distance between the unmanned vehicle and any of the preset navigation points is obtained based on the current position and the motion posture. If the minimum angle is less than or equal to a first threshold, the driving speed of the unmanned vehicle is adjusted based on the driving distance.

[0018] Furthermore, the motion map model calculates the optimal motion path including the following steps:

[0019] Using SLAM technology to combine the first image with map data, construct an environment map corresponding to the driving environment, and update the environment map based on the preset time interval to generate the motion map model;

[0020] The motion map model sets the current position as a starting point, sets the next destination as a target point, and generates a planned path between the starting point and the target point based on a dynamic window method;

[0021] The motion performance of the unmanned vehicle is obtained, the road surface parameters of the environment map are obtained, the planned path is modified based on the motion performance and the road surface parameters, and the modified planned path is set as the optimal motion path.

[0022] Further, modifying the planned path includes the following steps:

[0023] Mark traffic rules and risk roadblocks, determine the safety factor between the planned path and the risk roadblock based on the traffic rules, if the safety factor is less than a first preset value, insert multiple sub-target points between the target point and the starting point, generate a planned sub-path between the starting point and any sub-target point, and between any sub-target point and the target point based on the dynamic window method, repeat this step until the safety factor between any planned sub-path and the risk roadblock is greater than or equal to the first preset value, and combine all planned sub-paths into the modified planned path.

[0024] Furthermore, the process of completing the motion control of the unmanned vehicle includes the following steps:

[0025] The control unit obtains a driving instruction of the unmanned vehicle based on a feedback control algorithm, and adjusts the driving instruction in real time based on the image features, and the unmanned vehicle outputs the actual motion path based on the driving instruction;

[0026] The current position and the motion posture corresponding to the actual motion path within a preset time period are collected and set as historical data, a mathematical model is constructed based on the historical data, and a predicted motion trajectory of the unmanned vehicle in a future time period is predicted based on the mathematical model;

[0027] Constraints are set, and the predicted motion trajectory and the optimal motion path are fitted based on the constraints to generate a dynamically adjusted path. The feedback control algorithm dynamically adjusts the motion command based on the dynamically adjusted path, and the unmanned vehicle moves based on the dynamically adjusted motion command.

[0028] Furthermore, the constraint condition is set based on traffic rules and the movement performance of the unmanned vehicle.

[0029] Furthermore, the control target is defined according to the driving task and the environment map, the mathematical model includes the control target, the constraint conditions and the dynamic equation, the historical data is input into the mathematical model, and the dynamic equation is solved based on the numerical integration method to output the predicted motion trajectory.

[0030] Furthermore, the mathematical model is iteratively optimized based on the actual motion path.

[0031] Furthermore, the unmanned vehicle sets a plurality of marking points, sets the distance distribution and angle distribution between the unmanned aerial vehicle and the marking points, and sets the preset following mode based on the distance distribution and the angle distribution.

[0032] Compared with the prior art, the beneficial effects of the present invention are at least as follows:

[0033] The present invention first enables the UAV and the unmanned vehicle to move in coordination based on a preset following mode, which can improve the integrity of the first image set; then, image features are extracted through image preprocessing, and all image features are matched to accurately obtain the current position and motion posture of the unmanned vehicle; finally, the motion map model is constructed to generate the optimal motion path and corresponding motion instructions, which can quickly realize the motion control of the unmanned vehicle.

[0034] The present invention also monitors the actual motion path of the unmanned vehicle in real time by constructing a mathematical model, and can dynamically adjust the motion instructions in a timely manner to improve the safety and adaptability of the unmanned vehicle motion control. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 A flow chart of the steps of an unmanned vehicle motion control method based on regional visual relative positioning according to the present invention;

[0036] Figure 2 It is a schematic diagram of the preset following mode in the present invention. DETAILED DESCRIPTION

[0037] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0038] It is understood that the terms "first", "second", etc. used in this application may be used herein to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, without departing from the scope of this application, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script.

[0039] like Figure 1 As shown, a motion control method for an unmanned vehicle based on regional visual relative positioning includes:

[0040] S1: The drone moves in coordination with the unmanned vehicle based on a preset following mode, and the camera carried by the drone photographs the driving environment of the unmanned vehicle based on a preset time interval to generate a first image set.

[0041] Specifically, in this embodiment, the preset following mode is a method of coordinated movement, the camera includes but is not limited to a high-definition camera, the first image set refers to a set of first images taken by the camera, wherein the driving environment includes environmental information in all directions of the unmanned vehicle.

[0042] S2: Perform image preprocessing on any first image in the first image set, extract image features, and determine the current position and motion posture of the unmanned vehicle based on the image features.

[0043] Specifically, in this embodiment, image preprocessing refers to the process of processing the first image using image processing technology, and image features refer to key information in the first image, including but not limited to feature information such as road boundaries, obstacle locations, and traffic signs. The current position refers to the real-time position of the unmanned vehicle, and the motion posture refers to the direction and angle of the unmanned vehicle's motion.

[0044] S3: The control unit establishes a motion map model based on the current position and motion posture. The motion map model outputs an optimal motion path. Based on the optimal motion path, a motion instruction is sent to the unmanned vehicle. The unmanned vehicle moves after receiving the motion instruction.

[0045] Specifically, in this embodiment, the control unit is installed in the unmanned vehicle, and is used to generate motion instructions and control the motion mode of the unmanned vehicle. The motion map model refers to the environmental model of the unmanned vehicle when it is in motion generated by the regional visual relative positioning corresponding to the SLAM technology, which is used to simulate the change process of the driving environment. The optimal motion path refers to the recommended route for the unmanned vehicle to move, and the motion instruction refers to the operating instruction that controls the motion mode of the unmanned vehicle.

[0046] S4: Monitor the actual motion path of the unmanned vehicle in real time, dynamically adjust the motion instructions based on the actual motion path and the optimal motion path, and complete the motion control process of the unmanned vehicle.

[0047] Specifically, in this embodiment, the unmanned vehicle moves according to the motion instruction, and the actual motion path refers to the route track formed by the real-time change of the current position of the unmanned vehicle. The real-time monitoring of the actual motion path can determine whether the unmanned vehicle moves according to the motion instruction. The motion instruction is dynamically adjusted based on the actual motion path and the optimal motion path, so that the motion control process of the unmanned vehicle can be accurately and quickly realized.

[0048] As a preferred technical solution of the present invention, determining the current position and motion posture of the unmanned vehicle based on image features includes the following steps:

[0049] After the UAV acquires the regional image, it descends to a preset navigation point to photograph the driving environment. The first image set includes first images corresponding to the front, rear, left and right positions of the unmanned vehicle.

[0050] After image enhancement, feature extraction is performed on the first image to obtain image features, the image features are matched with preset navigation points to generate matching features, and the relative position of the unmanned vehicle is identified based on the matching features.

[0051] The actual position of the unmanned vehicle is calculated based on the flight data, matching features and relative position of the UAV, and the actual position is set as the current position.

[0052] The movement vector of any feature point group in the matching feature is obtained based on a preset time interval, all the movement vectors are fitted to generate a target vector, and the vector direction of the target vector is set as the motion posture.

[0053] Specifically, in this embodiment, the regional image refers to an image containing the entire unmanned vehicle. In order to obtain the first image at a fixed point, it is necessary to keep the drone and the unmanned vehicle at a relatively stable height, such as Figure 2 As shown, the drone T1 and the unmanned vehicle R1 maintain a relatively stable regional point and transmit data through the communication line L1, so that the drone can clearly capture the first image corresponding to the front, rear, left and right positions of the unmanned vehicle, wherein the preset navigation point refers to the navigation point of the path planning.

[0054] Image enhancement is a part of image preprocessing, which can improve the readability and recognizability of images. Image features are generated through feature extraction, where image features can refer to image areas composed of multiple feature points. The feature points contained in the image features are matched with the feature points contained in the preset navigation points through the SIFT (Scale-Invariant Feature Transform) algorithm, and the matched feature points are set as matching features. The corresponding preset navigation points can be identified by the matching features, thereby calculating the relative position of the unmanned vehicle, where the relative position refers to the position of the unmanned vehicle relative to the preset navigation point.

[0055] Flight data refers to the flight altitude, flight speed and other data of the drone, and matching features include spatial distance. Therefore, the actual position can be calculated by combining flight data, matching features and relative positions. Among them, the position of the preset navigation point is known, and the actual position refers to the actual position information of the unmanned vehicle in the map. Therefore, the actual position is set as the current position.

[0056] The feature points that match each other in the matching features are the feature point group. The displacement time of the feature point group can be determined by the preset time interval. The moving vector refers to the vector formed by the feature point group. Since there are multiple groups of feature point groups in the matching features, multiple moving vectors can be fitted to obtain the target vector, where the direction of the target vector is the vector direction, which can be used to describe the direction of change of the motion trajectory of the unmanned vehicle. Therefore, the vector direction is set as the motion posture.

[0057] As a preferred technical solution of the present invention, any two adjacent preset navigation points are connected based on the regional image to generate multiple connecting lines, the minimum angle between any two adjacent connecting lines is calculated, and the driving distance between the unmanned vehicle and any preset navigation point is obtained based on the current position and motion posture. If the minimum angle is less than or equal to a first threshold, the driving speed of the unmanned vehicle is adjusted based on the driving distance.

[0058] Specifically, in this embodiment, multiple preset navigation points are marked in the area image. The preset navigation points are usually set at forks in the road or locations where obstacles exist. The minimum angle can be used to determine whether any three adjacent preset navigation points are on the same straight line. Simply put, if the minimum angle is less than or equal to a threshold, it means that the route will turn when reaching the next preset navigation point. Therefore, the driving speed can be reduced based on the driving distance. At the same time, the minimum angle can also be used to control the rate of change of the motion posture to increase the safety of the unmanned vehicle movement.

[0059] As a preferred technical solution of the present invention, the motion map model calculates the optimal motion path including the following steps:

[0060] The first image is combined with the map data using SLAM technology to construct an environmental map corresponding to the driving environment, and the environmental map is updated based on a preset time interval to generate a motion map model.

[0061] The motion map model sets the current location as the starting point and the next destination as the target point, and generates a planned path between the starting point and the target point based on the dynamic window method.

[0062] The motion performance of the unmanned vehicle is obtained, the road surface parameters of the environment map are obtained, the planned path is modified based on the motion performance and the road surface parameters, and the modified planned path is set as the optimal motion path.

[0063] Specifically, in this embodiment, SLAM (Simultaneous Localization and Mapping) technology enables unmanned vehicles to navigate in unknown environments and build a map of the environment, i.e., an environmental map. Since unmanned vehicles are moving, the environmental map needs to be updated in real time. All environmental maps can be combined through machine learning to generate a motion map model.

[0064] Dynamic Window Approach (DWA) is a local motion planning algorithm for mobile robots or autonomous vehicles. It can search for the optimal motion instructions (such as driving speed and direction rotation rate) of unmanned vehicles in each control cycle to achieve obstacle avoidance and track predetermined paths and generate planned paths.

[0065] Motion performance refers to the performance parameters of the unmanned vehicle during its motion process, including but not limited to driving speed, direction rotation angle, and rotation rate. Road surface parameters refer to data such as the width of the drivable road surface and the shape of the route, which can be generated from the environmental map. Since the motion environment is changing, the planned path can be modified according to the road surface parameters and motion performance to generate the optimal motion path, thereby avoiding safety hazards in the motion of the unmanned vehicle when unexpected conditions occur on the road surface.

[0066] As a preferred technical solution of the present invention, modifying the planned path includes the following steps:

[0067] Mark traffic rules and risk roadblocks, determine the safety factor between the planned path and the risk roadblock based on the traffic rules, if the safety factor is less than a first preset value, insert multiple sub-target points between the target point and the starting point, generate a planned sub-path between the starting point and any sub-target point, and any sub-target point and the target point based on the dynamic window method, repeat this step until the safety factor between any planned sub-path and the risk roadblock is greater than or equal to the first preset value, and combine all planned sub-paths into a modified planned path.

[0068] Specifically, traffic rules refer to the rules and information such as traffic signs that unmanned vehicles need to comply with when moving. Risk roadblocks refer to existing roadblocks with potential safety hazards. The safety factor can be set by the ratio between the shortest distance between the planned path and the risk roadblock when passing through the risk roadblock and the road surface parameters. For example, when the shortest distance between the risk roadblock and the risk roadblock in the planned path is 1, and the value corresponding to the width in the road surface parameters is 4, its safety factor is 1 / 4. If the safety factor is less than the first preset value, it is necessary to divide the target point and the starting point into multiple sub-target points. The planned path can be refined through the generated planned sub-paths so that the safety factor corresponding to any planned sub-path is greater than or equal to the first preset value. The planned path can be modified according to the combined planned sub-paths.

[0069] As a preferred technical solution of the present invention, the motion control process of the unmanned vehicle includes the following steps:

[0070] The control unit obtains the driving instructions of the unmanned vehicle based on the feedback control algorithm, and adjusts the driving instructions in real time based on the image features. The unmanned vehicle outputs the actual motion path based on the driving instructions.

[0071] The current position and motion posture corresponding to the actual motion path within a preset time period are collected and set as historical data. A mathematical model is constructed based on the historical data, and the predicted motion trajectory of the unmanned vehicle in the future time period is predicted based on the mathematical model.

[0072] Constraints are set, and the predicted motion trajectory and the optimal motion path are fitted based on the constraints to generate a dynamically adjusted path. The feedback control algorithm dynamically adjusts the motion instructions based on the dynamically adjusted path, and the unmanned vehicle moves based on the dynamically adjusted motion instructions.

[0073] Specifically, in this embodiment, the feedback control algorithm is an algorithm for an automatic control system, which compares the actual output of the system with the expected output and adjusts the input of the system based on this error to make the actual output of the system as close to the expected output as possible, thereby adjusting the driving instructions in real time.

[0074] In order to monitor the actual movement path of the unmanned vehicle, it is necessary to collect data corresponding to the current position and movement posture within a preset time period, that is, historical data. The mathematical model can analyze and predict the historical data by constructing a dynamic equation, that is, output the predicted movement trajectory.

[0075] Constraints refer to conditional functions that restrict target variables in mathematical models, such as traffic rules, motion performance, etc. The dynamic adjustment path is generated in real time by fitting and comparing the predicted motion trajectory with the optimal motion path in combination with the mathematical model and the constraints. The dynamic adjustment path is used to dynamically adjust the motion instructions in real time, thereby controlling the motion of the unmanned vehicle, which can improve the accuracy and timeliness of the motion.

[0076] As a preferred technical solution of the present invention, constraint conditions are set based on traffic regulations and the motion performance of the unmanned vehicle.

[0077] Specifically, in this embodiment, traffic rules are the movement standards of unmanned vehicles, such as the setting of driving lanes and the traffic rules for semantic conversion after traffic sign recognition. Movement performance refers to performance parameters such as the movement speed of the unmanned vehicle. By setting constraints, the dynamic adjustment path can be accurately estimated.

[0078] As a preferred technical solution of the present invention, the control target is defined according to the driving task and the environmental map, the mathematical model includes the control target, constraints and dynamic equations, the historical data is input into the mathematical model, and the dynamic equations are solved based on the numerical integration method to output the predicted motion trajectory.

[0079] Specifically, the control target is the target variable of the mathematical model. Through the driving task and the environment map, the unmanned vehicle can move safely and accurately to the destination, which is the control target. The numerical integration method can improve the accuracy of the predicted motion trajectory by converting the continuous dynamic equation into a series of discrete steps to approximate the solution and checking whether the solution meets the constraints and accuracy requirements.

[0080] As a preferred technical solution of the present invention, the mathematical model is iteratively optimized based on the actual motion path.

[0081] Specifically, in this embodiment, the consistency between the actual motion path generated by the unmanned vehicle and the dynamically adjusted path is analyzed based on the timestamp, the accuracy of the mathematical model is evaluated, and then the mathematical model is iteratively optimized.

[0082] As a preferred technical solution of the present invention, the unmanned vehicle sets multiple marking points, sets the distance distribution and angle distribution between the unmanned vehicle and the marking points, and sets a preset following mode based on the distance distribution and angle distribution.

[0083] Specifically, in this embodiment, by constructing distance distribution and angle distribution through multiple marking points, it is possible to accurately determine whether the drone moves in coordination with the unmanned vehicle, thereby improving the usability of the first image set.

[0084] It should be understood that, although each step in the flow chart of each embodiment of the present invention is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0085] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the above-mentioned program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0086] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0087] The above embodiments only express several implementation methods of the present invention, and the description is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

[0088] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A motion control method for an unmanned vehicle based on regional visual relative positioning, characterized in that: The method comprises the following steps: S1: The drone moves in coordination with the unmanned vehicle based on a preset following mode, and the camera carried by the drone photographs the driving environment of the unmanned vehicle based on a preset time interval to generate a first image set; S2: performing image preprocessing on any first image in the first image set, extracting image features, and determining the current position and motion posture of the unmanned vehicle based on the image features; S3: The control unit establishes a motion map model based on the current position and the motion posture, the motion map model outputs an optimal motion path, sends a motion instruction to the unmanned vehicle based on the optimal motion path, and the unmanned vehicle moves after receiving the motion instruction; S4: monitoring the actual motion path of the unmanned vehicle in real time, and dynamically adjusting the motion instruction based on the actual motion path and the optimal motion path to complete the motion control process of the unmanned vehicle; The process of completing the motion control of the unmanned vehicle includes the following steps: The control unit obtains a driving instruction of the unmanned vehicle based on a feedback control algorithm, and adjusts the driving instruction in real time based on the image features, and the unmanned vehicle outputs the actual motion path based on the driving instruction; The current position and the motion posture corresponding to the actual motion path within a preset time period are collected and set as historical data, a mathematical model is constructed based on the historical data, and a predicted motion trajectory of the unmanned vehicle in a future time period is predicted based on the mathematical model; Constraints are set, and the predicted motion trajectory and the optimal motion path are fitted based on the constraints to generate a dynamically adjusted path. The feedback control algorithm dynamically adjusts the motion command based on the dynamically adjusted path, and the unmanned vehicle moves based on the dynamically adjusted motion command.

2. The method according to claim 1, characterized in that Determining the current position and motion posture of the unmanned vehicle based on the image features comprises the following steps: After acquiring the regional image, the drone descends to a preset navigation point to photograph the driving environment, wherein the first image set includes first images corresponding to the front, rear, left, and right directions of the unmanned vehicle respectively; The first image is subjected to feature extraction after image enhancement to obtain the image features, the image features are matched with the preset navigation points to generate matching features, and the relative position of the unmanned vehicle is identified based on the matching features; the computer vision technology calculates the actual position of the unmanned vehicle based on the flight data of the unmanned aerial vehicle, the matching features and the relative position, and sets the actual position as the current position; The movement vector of any feature point group in the matching feature is obtained based on the preset time interval, all the movement vectors are fitted to generate a target vector, and the vector direction of the target vector is set as the motion posture.

3. The method according to claim 2, characterized in that Based on the area image, any two adjacent preset navigation points are connected to generate multiple connecting lines, and the minimum angle between any two adjacent connecting lines is calculated. Based on the current position and the motion posture, the driving distance between the unmanned vehicle and any of the preset navigation points is obtained. If the minimum angle is less than or equal to a first threshold, the driving speed of the unmanned vehicle is adjusted based on the driving distance.

4. The method according to claim 1, characterized in that The motion map model calculates the optimal motion path including the following steps: Using SLAM technology to combine the first image with map data, construct an environment map corresponding to the driving environment, and update the environment map based on the preset time interval to generate the motion map model; The motion map model sets the current position as a starting point, sets the next destination as a target point, and generates a planned path between the starting point and the target point based on a dynamic window method; The motion performance of the unmanned vehicle is obtained, the road surface parameters of the environment map are obtained, the planned path is modified based on the motion performance and the road surface parameters, and the modified planned path is set as the optimal motion path.

5. The method according to claim 4, characterized in that Modifying the planned path includes the following steps: Mark traffic rules and risk roadblocks, determine the safety factor between the planned path and the risk roadblock based on the traffic rules, if the safety factor is less than a first preset value, insert multiple sub-target points between the target point and the starting point, generate a planned sub-path between the starting point and any sub-target point, and between any sub-target point and the target point based on the dynamic window method, repeat this step until the safety factor between any planned sub-path and the risk roadblock is greater than or equal to the first preset value, and combine all planned sub-paths into the modified planned path.

6. The method according to claim 1, characterized in that The constraint conditions are set based on traffic rules and the motion performance of the unmanned vehicle.

7. The method according to claim 1, characterized in that The control target is defined according to the driving task and the environment map. The mathematical model includes the control target, the constraint conditions and the dynamic equation. The historical data is input into the mathematical model, and the dynamic equation is solved based on the numerical integration method to output the predicted motion trajectory.

8. The method according to claim 1, characterized in that The mathematical model is iteratively optimized based on the actual motion path.

9. The method according to claim 1, characterized in that: The unmanned vehicle sets a plurality of marking points, sets the distance distribution and angle distribution between the unmanned aerial vehicle and the marking points, and sets the preset following mode based on the distance distribution and the angle distribution.

Images