Robot path planning method and system based on target attention point

Abstract

The invention belongs to the field of robot control, and provides a robot path planning method and system based on a target attention point, and the method comprises the steps: carrying out the environment field modeling based on a Gaussian process according to laser data obtained by a robot; based on a leading edge detection and clustering algorithm, generating target attention candidate points at known-unknown boundaries of an environment field, and taking prediction variances at the candidate points as measurement of field uncertainty; constructing a utility function fusing map information gain, field information gain and path cost, and selecting a target attention point from the candidate points by maximizing the utility function; and generating an optimal path by using a TEB path planner. According to the method, autonomous sampling of the robot can be effectively guided, information acquisition is maximized while the advancing path is optimized, and the precision and efficiency of environment field reconstruction are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of robot control, specifically relating to a robot path planning method and system based on target attention points. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] In typical scenarios such as indoor air quality assessment, marine observation, and pollutant dispersion analysis, environmental variables often exhibit highly non-uniform and time-varying distribution characteristics in unknown and structurally complex spaces. Relying on manual site deployment and inspection is not only costly and inefficient but also poses safety hazards. The development of mobile robot perception and navigation technologies provides a feasible approach for long-term autonomous environmental monitoring. However, under constraints of limited time and energy, how to rationally plan robot movement paths and sampling locations in unknown environments, while maximizing map coverage and minimizing uncertainties in environmental field reconstruction, is a key research problem in the field of environmental exploration and monitoring robots.

[0004] Currently, when robots simultaneously perform geometric mapping and environmental field reconstruction in unknown environments, there are still problems such as the separation of geometric and environmental field objectives, high computational overhead of information theory methods, and difficulty in parameter tuning for multi-objective trade-offs. Summary of the Invention

[0005] To address the aforementioned problems, this invention proposes a robot path planning method and system based on target attention points. This invention employs a global-level information-driven sampling decision and path planning strategy. Candidate sampling points are generated by clustering known and unknown region boundary grids, and attention target points with high information value are selected using an uncertainty gain index. During movement towards the attention target point, the robot collects observation data in real time and updates its Gaussian process regression model, achieving synchronous reconstruction of the environmental field. This effectively guides the robot's autonomous sampling, maximizing information acquisition while optimizing the travel path, significantly improving the accuracy and efficiency of environmental field reconstruction.

[0006] According to some embodiments, the present invention adopts the following technical solution: A robot path planning method based on target attention points includes the following steps: Based on the laser data acquired by the robot, environmental field modeling is performed using Gaussian processes; Based on frontier detection and clustering algorithms, target attention candidate points are generated at the known-unknown boundary of the environmental field, and the prediction variance at the candidate points is used as a measure of field uncertainty. Construct a utility function that integrates map information gain, field information gain, and path cost. By maximizing this utility function, select the target attention point from the candidate points. Use the TEB path planner to generate the optimal path.

[0007] As an alternative implementation, the process of modeling the environmental field based on Gaussian processes using laser data acquired by the robot includes: selecting a kernel function. For any unobserved location The posterior predictive distribution of environmental variables is:

[0008]

[0009] in, The covariance matrix between training samples, The covariance vector between the training samples and the predicted points. It is the covariance matrix between training points. The above predictions are all driven by observation data provided by the environmental field sensor array; Based on the posterior mean of environmental variables at various points in space This can generate an environmental field reconstruction map of the entire region, based on posterior variance. The uncertainty of the quantitative reconstruction results.

[0010] As an alternative implementation method, the process of generating target attention candidate points at the known-unknown boundary of the environmental field based on frontier detection and clustering algorithms includes: All frontier points in the currently occupied raster map are obtained through frontier detection, including the boundary raster between explored and unknown areas. Clustering algorithms are then used to group these frontier points, resulting in centroids for several candidate frontier regions. All centroids together form a candidate point set. .

[0011] As an alternative implementation, the process of using the prediction variance at candidate points as a measure of field uncertainty includes: defining the two-dimensional planar space to be explored as... The entire plane Everything was initially treated as unknown. = Areas that have already been explored Divided into and Based on the occupancy map constructed by laser SLAM, and using entropy To measure uncertainty in a map:

[0012] in It is a grid The probability of being occupied in a 2D SLAM occupancy map, where N represents the total number of grid cells in the entire occupancy map.

[0013] As a further step, to measure the performance of reducing map uncertainty, we use... Measuring the robot's performance in the candidate Entropy reduced by observation : = -

[0014] At the same time, query the field prediction variance at that point. As information gain for field reconstruction .

[0015] As an alternative implementation, constructing a utility function that integrates map information gain, field information gain, and path cost, and maximizing this utility function to select target attention points from candidate points includes calculating the target attention point utility index as follows:

[0016] in, Indicates the robot's current position With candidate points Path distance between and This represents the maximum information gain among all candidate points. In acquiring After the value, for For each candidate in the algorithm, path reachability and utility are evaluated. Path planning is performed from the robot's current position to the candidate point to determine if a feasible path exists. If no path exists, the candidate is marked as unreachable. For reachable candidates, calculate the path length or Euclidean distance. As a trade-off, the information gain of the candidates is normalized before calculating the target attention point utility index, and the maximum value is selected:

[0017] The candidate point corresponding to the maximum value is denoted as As the optimal target attention point at the global level.

[0018] As an alternative implementation, the following steps are also included: after generating the optimal path using the TEB path planner, during the robot's movement, the environmental field model is sampled at predetermined intervals and updated in real time through a Gaussian process, and iterated continuously until the task is completed.

[0019] As a further step, the process of sampling according to a predetermined period and updating the environmental field model in real time through a Gaussian process includes: tracking the robot's displacement relative to the previously sampled posture in real time, and when the accumulated Euclidean distance exceeds a preset threshold, recording new environmental readings and integrating them into the environmental field model.

[0020] A robot path planning system based on target attention points includes: The environmental field modeling module is configured to perform environmental field modeling based on Gaussian processes using laser data acquired by the robot. The candidate point generation module is configured to generate target attention candidate points at the known-unknown boundary of the environmental field based on the frontier detection and clustering algorithm, and use the prediction variance at the candidate points as a measure of field uncertainty. The target attention selection module is configured to construct a utility function that integrates map information gain, field information gain, and path cost. By maximizing this utility function, the target attention point is selected from the candidate points. The route planning module is configured to generate the optimal route using the TEB route planner.

[0021] A robot includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, perform the steps in the above-described method, or include the above-described system.

[0022] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention proposes a robot decision-making and planning method for environmental field exploration and reconstruction. It coordinates geometric map construction and environmental field modeling within a unified system, achieving integrated optimization of map coverage and field estimation accuracy.

[0023] This invention proposes a target attention point selection method that integrates occupancy map information gain and environmental field uncertainty measurement, compressing multi-source information into a single comprehensive utility index, which directly guides the robot to select the sampling location with the highest observation value.

[0024] The robot environment field exploration and reconstruction method proposed in this invention enables the planning results to be updated in real time with the environment field estimation, thereby improving the efficiency of environment field exploration and reconstruction accuracy under the constraint of limited motion cost.

[0025] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0026] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0027] Figure 1 This is a schematic diagram of the target attention point selection process in one embodiment; Figure 2 This is an example of environmental field and map settings; Figure 3 This is the result of random sampling environment reconstruction and corresponding path planning in one embodiment; Figure 4 This is an example of environment reconstruction and corresponding path planning results based on map information entropy; Figure 5 This is an example of environmental reconstruction and corresponding path planning results based on environmental field uncertainty; Figure 6 This is the result of environmental reconstruction and corresponding path planning based on the proposed method in one embodiment. Detailed Implementation

[0028] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0029] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, 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 pertains.

[0030] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0031] Where there is no conflict, the embodiments and features described in this application may be combined with each other.

[0032] Example 1 A robot path planning method based on target attention points includes the following steps: The robot constructs an occupation grid map using laser SLAM, builds an environmental field model based on Gaussian processes, and generates target attention candidate observation points at the known-unknown boundary based on frontier detection and clustering. The prediction variance at the candidate points is used as a measure of field uncertainty.

[0033] We further construct a utility function that integrates map information gain, field information gain, and path cost, and select the target attention point by maximizing this utility function.

[0034] The optimal path is generated using the TEB path planner. Regular sampling is performed during the movement process, and the environmental field model is updated in real time through a Gaussian process. Finally, efficient exploration and high-precision reconstruction of the unknown environment are achieved through iterative optimization.

[0035] The core problem that the above method aims to solve is the optimization of multi-objective driven exploration decisions in unknown environments. Specifically, the robot needs to select a global point with both high map information entropy and environmental field information entropy, enabling it to achieve high information gain in both exploring unknown regions and observing environmental field variables, while also considering path cost and reachability constraints. Let... For any candidate point, the set of candidate sampling points obtained by frontal clustering on the occupied grid map constructed for the geometric sensor is given. Define two types of information gain metrics: Map Information Gain The robot moves to the location. Subsequently, it measures the number of newly observed unknown grid cells or the reduction in map entropy within the sensor's field of view. This metric reflects the point's contribution to expanding the known environment.

[0036] Field reconstruction information gain Robots in The desired reduction in field uncertainty after measuring the environmental field. This can be approximated by the variance predicted by the current Gaussian process. This indicates that the larger the variance, the higher the potential value of the current position in improving the accuracy of the field model.

[0037] The main steps in this embodiment include: Environment field modeling based on Gaussian processes, observation target generation based on target attention points, and path execution and fixed step size sampling based on TEB.

[0038] First, environmental field modeling based on Gaussian processes: After selecting the kernel function, for any unobserved location The posterior predictive distribution of environmental variables (Gaussian process regression, also known as Kriging interpolation) is as follows:

[0039]

[0040] in The covariance matrix between training samples, The covariance vector between the training samples and the predicted points. It is the covariance matrix between training points. These predictions are all driven by observational data provided by the environmental field sensor array.

[0041] Based on the posterior mean of environmental variables at various points in space This allows the generation of a reconstructed environmental field map for the entire region. Posterior variance. This quantifies the uncertainty of the reconstruction results and provides crucial information for subsequent active sampling planning.

[0042] Secondly, target attention point generation: Given the aforementioned dual tasks, a wise trade-off needs to be made between the two objectives of "map exploration" and "environmental information collection".

[0043] This embodiment transforms the multidimensional objective function into a single comprehensive utility index. The process of selecting the target attention point is as follows: Figure 1 As shown in the figure. The solid curve represents the explored path, the gray area represents the unknown area, the diagonal area represents the boundary grid, the hollow circle represents the target attention candidate point, the solid circle represents the selected target attention candidate point, and the dashed curve represents the planned path to the target attention.

[0044] First, all leading edge points (boundary grids between explored and unknown regions) in the currently occupied raster map are obtained through leading edge detection. Then, a clustering algorithm is used to group the leading edge points, resulting in centroids for several candidate leading edge regions. All centroids form a candidate point set. .

[0045] Define the two-dimensional plane space to be explored as The entire plane Everything was initially treated as unknown. = Areas that have already been explored Divided into and To reduce the size of unknown regions, we base our approach on an occupied map constructed using SLAM, and use entropy... To measure uncertainty in a map:

[0046] in It is a grid The probability of occupancy in a 2D SLAM occupancy map, where N represents the total number of grid cells in the entire occupancy map. To measure the performance in reducing map uncertainty, this embodiment uses... To measure the robot's performance in the candidate Entropy reduced by observation : = -

[0047] At the same time, the field prediction variance at this point can be queried through the GP uncertainty service interface. As information gain for field reconstruction Approximate to .

[0048] because and There exists a trade-off relationship (e.g., some points can explore a large amount of new space but contribute little to the known field, and vice versa), making it difficult to evaluate the merits using a single scale. A decision criterion that balances both factors is needed. Furthermore, robot motion is subject to path constraints and dynamic constraints; the selected objective should be achievable while minimizing travel costs. Therefore, this patent reduces this problem to a comprehensive decision-making problem involving single-objective information gain optimization and path planning constraints. The proposed objective attention point utility index can be described as follows:

[0049] in Indicates the robot's current position With candidate points Path distance between and This represents the maximum information gain among all candidate points.

[0050] In acquiring After the value, for For each candidate, path reachability and utility are evaluated. The path planner is invoked to plan a path from the robot's current position to the candidate point, determining whether a feasible path exists. If no path exists, the candidate is marked as unreachable. For reachable candidates, the path length or Euclidean distance is calculated. As a trade-off, the information gain of the candidates is then normalized and then processed according to the aforementioned formula. Calculate its overall utility value. Finally, from... Select the maximum utility value:

[0051] The candidate point corresponding to the maximum value is denoted as As the optimal target attention point at the global level.

[0052] Finally, TEB-based path execution and fixed-step sampling: Determining the optimal attention target point Then, the navigation module will select the TEB path planner to go to the point. Unlike discrete sampling methods such as DWA, the TEB path planner treats the trajectory as a deformable band under optimization constraints. This method is more in line with the dynamic characteristics of the robot, so it has a better ability to dynamically plan the path to the target point, making the path smoother and more continuous.

[0053] Furthermore, to improve the temporal efficiency of the exploration, this embodiment proposes a continuous sampling protocol. The system tracks the robot's displacement relative to the previously sampled posture in real time. Only when the accumulated Euclidean distance exceeds a preset threshold (set to 1.0 meter in this embodiment) will a new environmental reading be recorded and integrated into the environmental field model. This method ensures that the training data is evenly distributed in space, thereby effectively avoiding data redundancy in low-speed scenarios, while also ensuring sufficient coverage during rapid transportation.

[0054] This embodiment also includes experimental setup and verification of the performance of the proposed method in simulation experiments.

[0055] To evaluate the performance of the proposed method, it was compared with the following representative methods in experiments: Random sampling path planner: Randomly selects a safe point on the map as the destination point to travel to.

[0056] Path planner based on map information entropy: entirely based on Select the global target point.

[0057] Path planner based on environmental uncertainties: entirely based on Select the global target point.

[0058] The accuracy of environmental field mapping is evaluated using two metrics: mean absolute error. With mean Kriging variance The definition is as follows: ; Where N represents the total number of locations to be estimated in the entire environmental field. 1 and 2 represent the estimated and actual values ​​at any location in the environmental field, respectively. The results are shown in Table 1.

[0059] Table 1: Results of the comparative experiment

[0060] As shown in Table 1, the proposed method demonstrates superior performance compared to other strategies across all evaluation metrics. Specifically, the method presented in this embodiment achieves the lowest average values ​​for both mean squared absolute error (MAE) and mean kriging variance (AKV), indicating that our algorithm ensures both accuracy and reliability in predicting the environmental field while simultaneously mapping the data.

[0061] This embodiment also constructs a 2D scene containing several obstacles in the Gazebo simulation environment, generates an occupied grid map using a LiDAR-based SLAM algorithm, and loads offline environmental ground truth data onto the grid at a fixed resolution. The differences in MAE, AKV, and path length are compared between the proposed method and baseline strategies (exploration based solely on map entropy, sampling based on GP variance, etc.). Experimental results are presented in... Figures 3-6 As shown in the image, in all experiments, the robot begins its exploration from the top left corner of the map.

[0062] As can be seen from the results above, the proposed method completes the exploration of the map and the accurate prediction of frequently exchanged information using the simplest path.

[0063] In summary, this embodiment constructs a unified target decision-making framework, generating target attention candidate points in the leading edge region and comprehensively considering the information gain and path cost brought by environmental observation to select the attention target point and its guiding path. By setting reasonable exploration termination conditions and performing Gaussian process prediction and result visualization on the entire map after task completion, this embodiment forms a complete technology chain from path planning, dual-sensor information fusion to univariate environmental field visualization reconstruction. The proposed method has good application prospects in scenarios such as environmental monitoring, pollutant diffusion assessment, and indoor / outdoor environmental cognition, and provides an effective technical solution for the active perception and environmental modeling of mobile robots.

[0064] Example 2 A robot path planning system based on target attention points includes: The environmental field modeling module is configured to perform environmental field modeling based on Gaussian processes using laser data acquired by the robot. The candidate point generation module is configured to generate target attention candidate points at the known-unknown boundary of the environmental field based on the frontier detection and clustering algorithm, and use the prediction variance at the candidate points as a measure of field uncertainty. The target attention selection module is configured to construct a utility function that integrates map information gain, field information gain, and path cost. By maximizing this utility function, the target attention point is selected from the candidate points. The route planning module is configured to generate the optimal route using the TEB route planner.

[0065] Example 3 A robot includes a memory and a processor, as well as computer instructions stored in the memory and running on the processor, which, when executed by the processor, perform the steps in the above-described method, or include the above-described system.

[0066] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of one or more computer-usable storage media (including, but not limited to, disk storage, etc.) containing computer-usable program code. CD - ROM It takes the form of a computer program product implemented on (such as optical memory, etc.).

[0067] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0068] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0069] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0070] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made by those skilled in the art without creative effort within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A robot path planning method based on target attention points, characterized in that, Includes the following steps: Based on the laser data acquired by the robot, environmental field modeling is performed using Gaussian processes; Based on frontier detection and clustering algorithms, target attention candidate points are generated at the known-unknown boundary of the environmental field, and the prediction variance at the candidate points is used as a measure of field uncertainty. Construct a utility function that integrates map information gain, field information gain, and path cost. By maximizing this utility function, select the target attention point from the candidate points. Use the TEB path planner to generate the optimal path.

2. The robot path planning method based on target attention points as described in claim 1, characterized in that, The process of modeling the environmental field based on Gaussian processes, using laser data acquired by the robot, includes: selecting a kernel function. For any unobserved location The posterior predictive distribution of environmental variables is: in, The covariance matrix between training samples, The covariance vector between the training samples and the predicted points. It is the covariance matrix between training points. The above predictions are all driven by observation data provided by the environmental field sensor array; Based on the posterior mean of environmental variables at various points in space This can generate an environmental field reconstruction map of the entire region, based on posterior variance. The uncertainty of the quantitative reconstruction results.

3. The robot path planning method based on target attention points as described in claim 1, characterized in that, The process of generating target attention candidate points at the known-unknown boundary of the environmental field based on frontier detection and clustering algorithms includes: All frontier points in the currently occupied raster map are obtained through frontier detection, including the boundary raster between explored and unknown areas. Clustering algorithms are then used to group these frontier points, resulting in centroids for several candidate frontier regions. All centroids together form a candidate point set. .

4. The robot path planning method based on target attention points as described in claim 1, characterized in that, The process of using the prediction variance at candidate points as a measure of field uncertainty includes: defining the two-dimensional plane space to be explored as... The entire plane Everything was initially treated as unknown. = Areas that have already been explored Divided into and Based on the occupancy map constructed by laser SLAM, and using entropy To measure uncertainty in a map: in It is a grid The probability of being occupied in a 2D SLAM occupancy map, where N represents the total number of grid cells in the entire occupancy map.

5. The robot path planning method based on target attention points as described in claim 4, characterized in that, To measure the performance of reducing map uncertainty, use Measuring the robot's performance in the candidate Entropy reduced by observation : = - At the same time, query the field prediction variance at that point. As information gain for field reconstruction .

6. The robot path planning method based on target attention points as described in claim 1, characterized in that, The process of constructing a utility function that integrates map information gain, field information gain, and path cost, and selecting the target attention point from candidate points by maximizing this utility function, includes calculating the utility index of the target attention point as follows: in, Indicates the robot's current position With candidate points Path distance between and This represents the maximum information gain among all candidate points. In acquiring After the value, for For each candidate in the algorithm, path reachability and utility are evaluated. Path planning is performed from the robot's current position to the candidate point to determine if a feasible path exists. If no path exists, the candidate is marked as unreachable. For reachable candidates, calculate the path length or Euclidean distance. As a trade-off, the information gain of the candidates is normalized before calculating the target attention point utility index, and the maximum value is selected: The candidate point corresponding to the maximum value is denoted as As the optimal target attention point at the global level.

7. The robot path planning method based on target attention points as described in claim 1, characterized in that, It also includes the following steps: After generating the optimal path using the TEB path planner, the environmental field model is sampled at predetermined intervals and updated in real time through a Gaussian process during the robot's movement, continuously iterating until the task is completed.

8. The robot path planning method based on target attention points as described in claim 7, characterized in that, The process of sampling according to a predetermined period and updating the environmental field model in real time through a Gaussian process includes: real-time tracking of the robot's displacement relative to the previously sampled posture; when the accumulated Euclidean distance exceeds a preset threshold, recording new environmental readings and integrating them into the environmental field model.

9. A robot path planning system based on target attention points, characterized in that, include: The environmental field modeling module is configured to perform environmental field modeling based on Gaussian processes using laser data acquired by the robot. The candidate point generation module is configured to generate target attention candidate points at the known-unknown boundary of the environmental field based on the frontier detection and clustering algorithm, and use the prediction variance at the candidate points as a measure of field uncertainty. The target attention selection module is configured to construct a utility function that integrates map information gain, field information gain, and path cost. By maximizing this utility function, the target attention point is selected from the candidate points. The route planning module is configured to generate the optimal route using the TEB route planner.

10. A robot comprising a memory and a processor, and computer instructions stored in the memory and running on the processor, wherein the computer instructions, when executed by the processor, perform the steps of the method of any one of claims 1-8, or comprise the system of claim 9.

Images