Robot autonomous following method based on single-line laser radar

Abstract

The invention relates to the technical field of robot autonomous following, in particular to a robot autonomous following method based on a single-line laser radar. Comprising the steps of collecting laser radar data, performing intelligent processing on the laser radar data to obtain accurate laser radar data, identifying contour features of a following target by using a simultaneous positioning and mapping technology, forming a reference point of the following target, and constructing a map; the constructed map is updated and optimized, and an optimized map is obtained; based on the optimized map and the accurate laser radar data, performing dynamic target tracking optimization and target motion prediction by using a dynamic adaptive path planning and target tracking algorithm to obtain a preliminary dynamic path plan; and based on the initial dynamic path planning, introducing a priority-based task scheduling mechanism to obtain the optimal path planning. The technical problems of low environmental adaptability and low tracing accuracy in the autonomous tracing implementation process of the robot in a closed environment are solved.

Description

Technical Field

[0001] The invention relates to the technical field of robot autonomous following, and in particular to a robot autonomous following method based on a single-line laser radar. Background Art

[0002] In the context of the rapid development of modern automation and intelligence, robotics technology has been widely used in many fields such as industry, services, and exploration. In particular, autonomous navigation technology in complex environments has become a hot topic in research and application. Autonomous navigation capability is one of the key technologies for robots to achieve high autonomy. It enables robots to perform tasks such as path planning, obstacle avoidance, and target tracking in unknown or partially unknown environments without human intervention. In autonomous navigation technology, lidar, as a high-precision sensor, is widely used in robot environmental perception and map construction because of its advantages such as long range, high accuracy, and no restrictions on lighting conditions. Through the environmental data obtained by lidar scanning, the robot can identify obstacles, paths, and other important features in the environment, providing important information for self-positioning and path planning.

[0003] However, existing technologies still have some limitations and challenges when facing complex and dynamically changing environments. The first is the problem of data accuracy. Although lidar has high accuracy, it is still interfered by factors such as noise and dynamic obstacles in actual applications, affecting the accuracy and reliability of the data. Secondly, the problem of dynamic environment adaptation. How to quickly and accurately identify and adapt to dynamically changing environments, such as avoiding moving obstacles, is still a technical problem. Thirdly, the problem of resource allocation and task scheduling. How to efficiently complete complex tasks under limited computing resources and energy conditions requires intelligent task scheduling and resource allocation strategies. In addition, issues such as dynamic target tracking and prediction, and path planning cost optimization are also key technical issues that need to be solved to achieve efficient autonomous navigation.

[0004] There are many methods for robot autonomous following. The Chinese invention patent "A robot autonomous following method based on single-line laser radar" proposed by Li Jinbo et al., application number: "CN202211682726.9", mainly includes: S1, obtain point cloud; S2, cluster segmentation; S3, target screening: filter out objects that are too large and too small through the upper and lower limits of the number of point clouds in the point cloud segmentation cluster; for the segmentation cluster obtained after filtering, calculate the fan-shaped area in front of the robot. If there is no target in the front fan-shaped area, calculate the targets on both sides of the robot; then compare the target coordinates of the robot at the current moment with the target coordinates of the previous moment in the robot coordinate system. If the target distance between the two moments is greater than the predetermined distance value, filter out this target; S4, calculate the control speed; S5, follow the target. On the one hand, the present invention has small size, low cost, low computing power requirements, strong anti-interference to sunlight, and good generalization ability; on the other hand, the calculation amount and memory occupancy rate are small, the process is relatively simple, easy to debug, and the effect achieved has met the needs and reduced the probability of false detection.

[0005] However, in the process of implementing the technical solution of the invention in the above application embodiment, the inventors of the present application found that the above technology has at least the following technical problems: low environmental adaptability and low tracking accuracy during the robot's autonomous tracking in a closed environment. Summary of the invention

[0006] A robot autonomous following method based on a single-line laser radar comprises the following steps:

[0007] S1. Collect LiDAR data, perform intelligent processing on the LiDAR data, and obtain accurate LiDAR data; Based on the accurate LiDAR data, use the simultaneous positioning and mapping technology to identify the contour features of the target to be followed, form the reference point of the target to follow, and realize the robot's self-positioning and map construction in the working environment; update and optimize the constructed map to obtain an optimized map;

[0008] S2. Based on the optimized map and accurate lidar data, dynamic adaptive path planning and target tracking algorithm are used to perform dynamic target tracking optimization and target motion prediction to obtain preliminary dynamic path planning; based on the preliminary dynamic path planning, a priority-based task scheduling mechanism is introduced to obtain the optimal path planning.

[0009] Preferably, the S1 specifically includes:

[0010] The light and shadow navigation algorithm is introduced to analyze the light intensity distribution and shadow changes in the accurate lidar data, and intelligently identify and filter the noise introduced by dynamic obstacles.

[0011] Preferably, the S1 specifically includes:

[0012] In the process of implementing the light and shadow navigation algorithm, the inter-frame light intensity change is calculated and normalized to obtain the normalized inter-frame light intensity change; based on the normalized inter-frame light intensity change, a threshold is set to determine whether the scanning point is affected by a dynamic object. The specific implementation is as follows:

[0013]

[0014] If ΔI norm (i) exceeds the threshold value θ ΔI , then the scanning point i is considered to be in the shadow of a dynamic object, that is, IsDynαmic(i) = True; otherwise, if ΔI norm (i) Does not exceed the threshold θ ΔI , it is considered that the scanning point i is not in the shadow of the dynamic object, that is, IsDynamic(i) = False; IsDynamic(i) represents the state of the dynamic object; ΔI norm (i) is the normalized inter-frame light intensity variation at scanning point i.

[0015] Preferably, the S1 specifically includes:

[0016] In the process of implementing the light and shadow navigation algorithm, a dynamic feature mask is constructed to mark the scanning points in the shadows of dynamic objects. Based on the dynamic feature masks of all scanning points, a dynamic feature mask set is obtained, and based on the dynamic feature mask set, the scanning points affected by dynamic objects are filtered.

[0017] Preferably, the S1 specifically includes:

[0018] A spatiotemporal mapping algorithm is introduced to update map information by utilizing the spatiotemporal correlation between historical lidar data and current lidar data; map details are adjusted by comparing changes in environmental features over time series.

[0019] Preferably, the S1 specifically includes:

[0020] In the process of implementing the spatiotemporal mapping algorithm, the lidar data at continuous time points are collected, a time series is established for each feature point, and the change trend of the feature point is analyzed; and the exponential smoothing method is applied to the time series data of each feature point to calculate the change trend. The specific implementation is as follows:

[0021] S t =αx t +(1-α)S t-1

[0022] Among them, S t is the estimated value of the feature point at time t; x t is the actual observation value of the lidar data at time t; St-1 is the estimated value of the feature point at time t-1; α is the smoothing coefficient.

[0023] Preferably, the S1 specifically includes:

[0024] In the implementation process of the spatiotemporal mapping algorithm, the time series of each feature point is analyzed, the difference of the feature point estimation value is calculated, and the change of the feature point estimation value is obtained; when the change of the feature point estimation value exceeds the threshold, the map is updated and optimized to obtain the optimized map.

[0025] Preferably, the S2 specifically includes:

[0026] In introducing the priority-based task scheduling mechanism, the tasks performed by the robot in the working environment are identified, and a priority function is introduced to assign a priority to each task. The priority function:

[0027]

[0028] Among them, P j is the priority of task j; U j is the urgency score; Γ j is the importance score; D j is the deadline of task j; C t is the current time; R j is the amount of resources required to complete task j; is the adjustment factor.

[0029] Preferably, the S2 specifically includes:

[0030] In the resource allocation process, resource evaluation is performed to obtain a total resource vector; based on the task priority and the total resource vector, resources are allocated to each task using a resource allocation model to obtain a resource allocation result.

[0031] Preferably, the S2 specifically includes:

[0032] Based on the resource allocation result, a path cost function is used to perform dynamic path planning for each task; and an optimization algorithm is used to optimize the dynamic path planning to obtain the best path planning; the path cost function is:

[0033]

[0034] Among them, C path is the total cost of the path; N is the total number of tasks; L j is the path length of task j; is the estimated time for task j; A j is the amount of resources allocated to task j; S j is the smoothness of the path of task j; Ej is the estimated energy consumption of task j; is the weight factor.

[0035] The beneficial effects of the technical solution of the present invention are:

[0036] 1. Through intelligent processing of LiDAR data, including filtering, denoising and anomaly detection, the accuracy of the data is greatly improved, providing a more reliable foundation for subsequent target recognition, self-positioning and map construction, ensuring that the robot's perception of the environment is more accurate, thereby improving the overall performance of the navigation system; the introduction of light and shadow navigation algorithm analyzes light intensity changes to identify dynamic objects, effectively reducing the interference of dynamic obstacles on environmental feature extraction, so that the robot can stably perform self-positioning and path planning even in a dynamically changing environment, improving the reliability and safety of navigation; the introduction of spatiotemporal mapping algorithm utilizes the spatiotemporal correlation between historical and current LiDAR data to dynamically update map information, so that the map can reflect the latest status of the environment in a timely manner, ensuring the accuracy of the robot map, and enabling the navigation system to quickly adapt to changes in the environment; the simultaneous positioning and mapping technology is used to identify the position and contour information of the follow-up target, realizing the robot's self-positioning and map construction in the working environment.

[0037] 2. The priority-based task scheduling mechanism ensures that urgent and important tasks are executed first, and the task execution order is intelligently adjusted according to available resources, which optimizes resource allocation and improves the speed and quality of task completion. Through clustering algorithms and Kalman filtering, the robot can track and predict the position of dynamic targets more accurately, providing precise input for dynamic path planning. The introduced path cost function comprehensively considers factors such as path length, estimated time, resource quantity, path smoothness and energy consumption, making dynamic path planning both efficient and energy-saving. In particular, the dynamic path planning is further optimized through the optimization algorithm, ensuring that the found path is the most cost-effective, thereby reducing the overall cost of robot operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a flow chart of the robot autonomous following method based on single-line laser radar described in this application. DETAILED DESCRIPTION

[0039] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0040] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0041] The specific scheme of the robot autonomous following method based on a single-line laser radar provided by the present invention is described in detail below with reference to the accompanying drawings.

[0042] See attached Figure 1 , which shows a structure diagram of a robot autonomous following method based on a single-line laser radar provided by an embodiment of the present invention, comprising the following steps:

[0043] S1. Collect LiDAR data, perform intelligent processing on the LiDAR data, and obtain accurate LiDAR data; Based on the accurate LiDAR data, use the simultaneous positioning and mapping technology to identify the contour features of the target to be followed, form the reference point of the target to follow, and realize the robot's self-positioning and map construction in the working environment; update and optimize the constructed map to obtain an optimized map;

[0044] Use laser radar to scan the environment to obtain laser radar data, and perform intelligent processing on the laser radar data to obtain accurate laser radar data. The intelligent processing specifically includes preprocessing the acquired laser radar data such as filtering and denoising to improve the accuracy and availability of the laser radar data to obtain preprocessed laser radar data; perform anomaly detection processing on the preprocessed laser radar data to obtain laser radar data after anomaly detection, that is, accurate laser radar data;

[0045] Based on accurate LiDAR data, the robot uses simultaneous positioning and mapping technology to identify the contour features of the target to be followed, form the reference points of the target to follow, and realize the robot's self-positioning and map construction in the working environment. The specific implementation process is as follows:

[0046] The gradient method is used to identify the features in the environment with accurate LiDAR data to obtain environmental features. The environmental features, such as corners and furniture edges, provide anchor points for subsequent positioning and map construction. An existing trained deep learning model is used to identify following targets of specific shapes and sizes. The input of the deep learning model is accurate LiDAR data, and the output is the position and contour information of the following target. Then, the SLAM algorithm is used to realize the real-time positioning of the robot and the construction of the environmental map according to the extracted environmental features and the position and contour information of the following target. At the same time, during the movement of the robot, the map information is updated in real time, and the accuracy of the map is optimized to adapt to the dynamic changes of the environment. Based on the contour information of the following target, the geometric center of the following target is selected as the reference point as a following reference to realize the self-positioning and map construction of the robot in the working environment.

[0047] In order to avoid interference from dynamic obstacles when extracting environmental features in a dynamic environment, the light and shadow navigation algorithm is introduced. The light and shadow navigation algorithm analyzes the light intensity distribution and shadow changes in accurate lidar data, intelligently identifies and filters the noise introduced by dynamic obstacles, and improves the accuracy of environmental feature extraction. The core of the light and shadow navigation algorithm is to use the continuity of light intensity changes and the regularity of spatial distribution to distinguish static objects from dynamic objects; the specific implementation process is as follows:

[0048] First, calculate the light intensity change between frames; for two consecutive scanning frames t and t-1, calculate the light intensity change of each scanning point i between the two frames; record I t (i) and I t -1(i) are the light intensity values ​​at scanning point i at time t and t-1 respectively;

[0049] ΔI(i)=|I t (i)-I t-1 (i)|

[0050] Where ΔI(i) is the change in light intensity between frames; the light intensity value can be represented by the radar reflection intensity, and the change in radar reflection intensity reflects the change in the surface characteristics or position of the object;

[0051] Then normalize the light intensity change between frames; in order to compare the light intensity change at different scanning points, normalize ΔI(i):

[0052]

[0053] Among them, ΔI norm (i) is the normalized inter-frame light intensity variation at scanning point i. Through the normalization process, the inter-frame light intensity variation of each scanning point is normalized to the range between 0 and 1, which is convenient for subsequent processing.

[0054] Furthermore, a threshold θ is set by expert experience. ΔI , to identify significant frame-to-frame light intensity changes, indicating the presence of dynamic objects;

[0055]

[0056] If ΔI norm (i) exceeds the threshold value θ ΔI , then the scanning point i is considered to be in the shadow of a dynamic object, that is, IsDynamic(i) = True; otherwise, if ΔI norm (i) Does not exceed the threshold θ ΔI , it is considered that the scanning point i is not in the shadow created by the dynamic object, that is, IsDynamic(i) = False; IsDynamic(i) indicates the state of the dynamic object;

[0057] Further, after determining the scan points affected by the dynamic object, the scan points are filtered to reduce interference with subsequent SLAM processing;

[0058] Reconstruct the dynamic feature mask; According to the result of IsDynamic(i), construct a dynamic feature mask M for each scanning point dynamic (i)

[0059]

[0060] The dynamic feature mask will be used to mark those scan points that may be affected by dynamic objects;

[0061] Based on the dynamic feature masks of all scanning points, the dynamic feature mask set M is obtained dynamic , and based on the dynamic feature mask set M dynamic Filter out scan points affected by dynamic objects;

[0062] x filtered (i) = x(i)·(1-M dynamic (i))

[0063] Among them, x(i) is the coordinate value of the scan point that is not filtered and affected by the dynamic object, x filtered (i) is the coordinate value of the scanned point after filtering, which is used for subsequent SLAM processing;

[0064] The above process effectively eliminates the interference of dynamic objects from the accurate lidar data, providing more accurate static environment information for SLAM;

[0065] In order to avoid the map becoming outdated due to environmental changes over time, a spatiotemporal mapping algorithm is introduced. The spatiotemporal mapping algorithm uses the spatiotemporal association between historical lidar data and current lidar data to dynamically update map information so that the map can reflect the latest state of the environment. By comparing the changes in environmental characteristics over time series, the map details can be intelligently adjusted. The specific implementation process is as follows:

[0066] First, collect the lidar data at consecutive time points and establish a time series for each feature point in the environment; then analyze the trend of feature point changes; apply exponential smoothing method to the time series data of each feature point to calculate the trend of changes;

[0067] S t =αx t +(1-α)S t-1

[0068] Among them, S t is the estimated value of the feature point at time t; xt is the actual observation value of the lidar data at time t; S t-1 is the feature point estimate at time t-1; α is the smoothing coefficient, which determines the degree of influence of historical lidar data on the current estimate, 0<α<1;

[0069] Analyze the time series of each feature point and identify significant changes by calculating the difference of the feature point estimates at consecutive moments:

[0070] ΔS=|S t -S t-1 |

[0071] Wherein, ΔS is the change in the estimated value of the feature point from time t-1 to time t; the feature points with significant changes are determined by setting a threshold δ according to the expert experience method and updated on the map; that is, if the change ΔS of the estimated value of a feature point exceeds the threshold δ (that is, ΔS>δ), the feature point is considered to have changed significantly, and a significant change means that the state of the feature point in the environment (such as the state such as position, shape or other attributes) has changed significantly, and the map needs to be updated in the SLAM process to reflect the new state;

[0072] During the updating process, known environmental features and newly observed significant changes are utilized, and spatial analysis methods are used to optimize the overall accuracy and continuity of the map to obtain an optimized map.

[0073] S2. Based on the optimized map and accurate LiDAR data, the dynamic adaptive path planning and target tracking algorithm are used to optimize dynamic target tracking and predict target motion to obtain preliminary dynamic path planning. Based on the preliminary dynamic path planning, a priority-based task scheduling mechanism is introduced to obtain the optimal path planning.

[0074] Based on the optimized map and accurate LiDAR data, dynamic adaptive path planning and target tracking algorithms are used to optimize dynamic target tracking and predict target motion to achieve dynamic path planning. The specific implementation process is as follows:

[0075] First, based on accurate LiDAR data, a clustering algorithm is used to identify the position of the target, and a Kalman filter is applied to each target for motion prediction to obtain the predicted position of the target. Then, based on the optimized map and the predicted position of the target, the A* algorithm is used to calculate the optimal path from the current position to the target position. The path found by the A* algorithm is further smoothed using a Bezier curve to adapt to the robot's motion characteristics, generate a more practical and feasible path, and obtain preliminary dynamic path planning.

[0076] Based on the preliminary dynamic path planning, a priority-based task scheduling mechanism is introduced to optimize the execution order and resource allocation, obtain the best path planning, and realize the robot's autonomous tracking; the specific implementation process is as follows:

[0077] First, task identification and priority allocation; for task identification, existing recognition technology is used to identify the tasks that the robot needs to perform in the working environment, including dynamic target tracking, obstacle avoidance, and path finding;

[0078] Priority is assigned to the identified tasks, and a priority function is used to assign a priority to each task according to the urgency and importance of the task. The priority function is:

[0079]

[0080] Among them, P j is the priority of task j; U j is the urgency score, ranging from [0, 1], where 1 represents the most urgent; Γ j is the importance score, and its value range is also [0, 1], where 1 represents the most important; D j is the deadline of task j; C t is the current time; R j is the amount of resources required to complete task j; It is an adjustment factor, which is used to adjust the weight of different factors on the priority;

[0081] Use the above process as a priority task scheduling mechanism;

[0082] For resource allocation, we first conduct resource evaluation and evaluate the available resources based on expert experience to obtain the total resource vector: R = [r1, r2, ..., r μ ,…,r n ] T ; where n represents the number of resource categories; r μ represents the available amount of the μth resource, μ∈[1,n]; the available resources include computing power, power, sensor bandwidth, etc.;

[0083] According to the task priority and the total resource vector, resources are allocated to each task using a resource allocation model, and the resource allocation model is specifically as follows:

[0084] A j =f(P j , R)

[0085]

[0086] Among them, A j is the resource vector assigned to task j; f is the resource allocation function, according to the priority P of the taskj and the total resource vector R to determine resource allocation; N is the total number of tasks; k is a positive coefficient used to adjust the priority P of the task j sensitivity to the impact on resource allocation;

[0087] Further, dynamic path planning optimization is performed; according to the resource allocation result, dynamic path planning is performed for each task using a path cost function; the path cost function:

[0088]

[0089] Among them, C path is the total cost of the path; L j is the path length of task j; is the estimated time for task j; A j is the amount of resources allocated to task j; S j is the smoothness of the path of task j; E j is the estimated energy consumption of task j; is a weight factor that reflects the different preferences of different tasks for path length, estimated time, amount of resources, path smoothness, and energy consumption;

[0090] Finally, an optimization algorithm (such as genetic algorithm) is used to optimize dynamic path planning, with minimizing the total cost of the path as the optimization goal. The optimization algorithm is implemented to find the path that can minimize the total cost of the path, that is, the optimal path planning, to achieve autonomous tracking of the robot.

[0091] In summary, a robot autonomous following method based on single-line laser radar was completed.

[0092] The order of the embodiments of the invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0093] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0094] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. A robot autonomous following method based on a single-line laser radar, characterized in that: The following steps are involved: S1. Collect LiDAR data, perform intelligent processing on the LiDAR data, and obtain accurate LiDAR data; Based on the accurate LiDAR data, use the simultaneous positioning and mapping technology to identify the contour features of the target to be followed, form the reference point of the target to follow, and realize the robot's self-positioning and map construction in the working environment; update and optimize the constructed map to obtain an optimized map; S2. Based on the optimized map and accurate lidar data, dynamic adaptive path planning and target tracking algorithm are used to perform dynamic target tracking optimization and target motion prediction to obtain preliminary dynamic path planning; based on the preliminary dynamic path planning, a priority-based task scheduling mechanism is introduced to obtain the optimal path planning.

2. The robot autonomous following method based on a single-line laser radar according to claim 1 is characterized in that: The S1 specifically includes: The light and shadow navigation algorithm is introduced to analyze the light intensity distribution and shadow changes in the accurate lidar data, and intelligently identify and filter the noise introduced by dynamic obstacles.

3. The robot autonomous following method based on a single-line laser radar according to claim 2 is characterized in that: The S1 specifically includes: In the process of implementing the light and shadow navigation algorithm, the inter-frame light intensity change is calculated and normalized to obtain the normalized inter-frame light intensity change; based on the normalized inter-frame light intensity change, a threshold is set to determine whether the scanning point is affected by a dynamic object. The specific implementation is as follows: If ΔI norm (i) exceeds the threshold value θ ΔI , then the scanning point i is considered to be in the shadow of a dynamic object, that is, IsDynamic(i) = True; otherwise, if ΔI norm (i) Does not exceed the threshold θ ΔI , it is considered that the scanning point i is not in the shadow of the dynamic object, that is, IsDynamic(i) = False; IsDynamic(i) represents the state of the dynamic object; ΔI norm (i) is the normalized inter-frame light intensity variation at scanning point i.

4. The robot autonomous following method based on a single-line laser radar according to claim 3 is characterized in that: The S1 specifically includes: In the process of implementing the light and shadow navigation algorithm, a dynamic feature mask is constructed to mark the scanning points in the shadows of dynamic objects. Based on the dynamic feature masks of all scanning points, a dynamic feature mask set is obtained, and based on the dynamic feature mask set, the scanning points affected by dynamic objects are filtered.

5. The robot autonomous following method based on single-line laser radar according to claim 1 is characterized in that: The S1 specifically includes: A spatiotemporal mapping algorithm is introduced to update map information by utilizing the spatiotemporal correlation between historical lidar data and current lidar data; map details are adjusted by comparing changes in environmental features over time series.

6. The robot autonomous following method based on single-line laser radar according to claim 5, characterized in that: The S1 specifically includes: In the process of implementing the spatiotemporal mapping algorithm, the lidar data at continuous time points are collected, a time series is established for each feature point, and the change trend of the feature point is analyzed; and the exponential smoothing method is applied to the time series data of each feature point to calculate the change trend. The specific implementation is as follows: S t =αx t +(1-α)S t-1 Among them, S t is the estimated value of the feature point at time t; x t is the actual observation value of the lidar data at time t; S t-1 is the estimated value of the feature point at time t-1; α is the smoothing coefficient.

7. The robot autonomous following method based on single-line laser radar according to claim 6 is characterized in that: The S1 specifically includes: In the implementation process of the spatiotemporal mapping algorithm, the time series of each feature point is analyzed, the difference of the feature point estimation value is calculated, and the change of the feature point estimation value is obtained; when the change of the feature point estimation value exceeds the threshold, the map is updated and optimized to obtain the optimized map.

8. The robot autonomous following method based on single-line laser radar according to claim 1, characterized in that: The S2 specifically includes: In introducing the priority-based task scheduling mechanism, the tasks performed by the robot in the working environment are identified, and a priority function is introduced to assign a priority to each task. The priority function: Among them, P j is the priority of task j; U j is the urgency score; Γ j is the importance score; D j is the deadline of task j; C t is the current time; R j is the amount of resources required to complete task j; is the adjustment factor.

9. The robot autonomous following method based on single-line laser radar according to claim 8, characterized in that: The S2 specifically includes: In the resource allocation process, resource evaluation is performed to obtain a total resource vector; based on the task priority and the total resource vector, resources are allocated to each task using a resource allocation model to obtain a resource allocation result.

10. The robot autonomous following method based on single-line laser radar according to claim 9, characterized in that: The S2 specifically includes: Based on the resource allocation result, a path cost function is used to perform dynamic path planning for each task; and an optimization algorithm is used to optimize the dynamic path planning to obtain the best path planning; the path cost function is: Among them, C path is the total cost of the path; N is the total number of tasks; L j is the path length of task j; is the estimated time for task j; A j is the amount of resources allocated to task j; S j is the smoothness of the path of task j; E j is the estimated energy consumption of task j; is the weight factor.

Images