CN-121977591-A - Robot path planning method and system based on target attention point

CN121977591ACN 121977591 ACN121977591 ACN 121977591ACN-121977591-A

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, wherein environmental field modeling is performed based on a Gaussian process according to laser data acquired by a robot; generating target attention points at known-unknown boundaries of an environmental field based on a front edge detection and clustering algorithm, taking prediction variance at the candidate points as a measure of field uncertainty, constructing a utility function integrating map information gain, field information gain and path cost, selecting the target attention points from the candidate points by maximizing the utility function, and generating an optimal path by using a TEB path planner. The invention can effectively guide the robot to autonomously sample, optimize the travelling path and simultaneously maximize the information acquisition, thereby remarkably improving the accuracy and efficiency of the reconstruction of the environmental field.

Inventors

LI TENG
LIU SHUPENG
CAO KAIQI
WANG YUNXIAO

Assignees

山东大学

Dates

Publication Date: 20260505
Application Date: 20251223

Claims (10)

1. The robot path planning method based on the target attention point is characterized by comprising the following steps of: according to laser data acquired by the robot, modeling an environmental field based on a Gaussian process; generating target attention candidate points at known-unknown boundaries of the environmental field based on a leading edge detection and clustering algorithm, and taking prediction variances at the candidate points as measures of field uncertainty; constructing a utility function integrating the map information gain, the field information gain and the path cost, and selecting a target attention point from candidate points by maximizing the utility function; An optimal path is generated using a TEB path planner.
2. A robot path planning method based on target attention point according to claim 1, wherein the process of modeling the environment field based on Gaussian process based on laser data acquired by the robot comprises selecting kernel functions For any unobserved position The posterior predictive distribution of the environmental variables is: Wherein, the To train the covariance matrix between the samples, To train the covariance vector between the samples and the predicted points, Is the covariance matrix between the training points, The prediction quantity is driven by observation data provided by an environmental field sensor group; posterior mean value at each point in space based on environment variable An environment field reconstruction map of the whole area can be generated based on the posterior variance And quantifying uncertainty of the reconstruction result.
3. A method of planning a path for a robot based on a target point of attention as recited in claim 1, wherein generating a target point of attention candidate at a known-unknown boundary of an environmental field based on a leading edge detection and clustering algorithm comprises: acquiring all front points in the current occupied grid map through front edge detection, wherein the front edge points comprise boundary grids of explored areas and unknown areas, clustering the front edge points by using a clustering algorithm to obtain centroids of a plurality of front edge candidate areas, and forming a candidate point set by all centroids 。
4. A robot path planning method based on target points of attention as claimed in claim 1, wherein the process of taking the predicted variance at the candidate points as a measure of field uncertainty comprises defining the two-dimensional planar space to be explored as The whole plane All at the beginning are treated as unknown = Areas that have been explored Is divided into And Based on occupancy maps constructed by laser SLAM, entropy is used To measure uncertainty in the map: Wherein the method comprises the steps of Is a grid Probability of occupancy in a two-dimensional SLAM occupancy map, N represents the total number of grids in the overall occupancy map.
5. The method for planning a path of a robot based on a target point of attention as recited in claim 4, wherein for measuring performance of reducing map uncertainty, using Measuring robot presence candidates Entropy reduced by observation : = - At the same time, the field prediction variance of the point is queried Gain as field reconstruction information 。
6. The robot path planning method according to claim 1, wherein constructing a utility function that merges the map information gain, the field information gain and the path cost, and selecting the target point of attention from the candidate points by maximizing the utility function comprises calculating a target point of attention utility index as: Wherein, the Representing the current position of a robot And candidate points The distance of the path between them, And Maximum value of gain of each information in all candidate points; During acquisition of After the value, pair Carrying out path accessibility and utility evaluation on each candidate in the robot, carrying out path planning on the current position of the robot to candidate points, and judging whether a feasible path exists or not; for reachable candidates, the path length or Euclidean distance is calculated As a cost, calculating a target attention point utility index after normalizing the candidate information gain, and selecting a maximum value: The candidate point corresponding to the maximum value is marked as As the optimal target point of attention for the global level.
7. The method for planning a path of a robot based on a target point of attention as recited in claim 1, further comprising the steps of sampling and updating the environmental field model in real time through a Gaussian process in a predetermined period during movement of the robot after generating an optimal path by using the TEB path planner, and iterating until a task is completed.
8. A method for planning a path for a robot based on a target point of attention as recited in claim 7, wherein sampling at a predetermined period and updating the environmental field model in real time by a Gaussian process includes tracking the displacement of the robot relative to a previously sampled pose in real time, recording a new environmental reading and integrating it into the environmental field model when the accumulated Euclidean distance exceeds a preset threshold.
9. A robot path planning system based on a target point of attention, comprising: the environment field modeling module is configured to perform environment field modeling based on a Gaussian process according to laser data acquired by the robot; A candidate point generation module configured to generate target attention candidate points at known-unknown boundaries of the environmental field based on a leading edge detection and clustering algorithm, taking a prediction variance at the candidate points as a measure of field uncertainty; the target attention selecting module is configured to construct a utility function integrating the map information gain, the field information gain and the path cost, and select a target attention point from the candidate points by maximizing the utility function; And a path planning module configured to generate an optimal path using the TEB path planner.
10. A robot comprising a memory and a processor and computer instructions stored on the memory and running on the processor, which when executed by the processor, perform the steps in the method of any one of claims 1-8, or comprise the system of claim 9.

Description

Robot path planning method and system based on target attention point Technical Field The invention belongs to the field of robot control, and particularly relates to a robot path planning method and system based on a target attention point. Background The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art. In typical scenarios such as indoor air quality assessment, marine observation and contaminant diffusion analysis, environmental variables tend to exhibit highly non-uniform and time-varying distribution characteristics in an unknown and structurally complex space. The manual distribution and inspection are relied on, so that the cost is high, the efficiency is low, and potential safety hazards exist. The development of mobile robot sensing and navigation technology provides a viable approach to long-term autonomous environmental monitoring. However, under the constraint of finite time and energy, how to reasonably plan the motion path and the sampling position of the robot in an unknown environment, and the map coverage rate is improved while the uncertainty of the reconstruction of an environment field is reduced to the greatest extent, so that the method is a key research problem in the field of environment exploration and monitoring robots. At present, when the robot completes geometric map construction and environment field reconstruction in an unknown environment at the same time, the problems of target fracture of geometric and environment fields, high calculation cost of an information theory method, difficulty in parameter adjustment due to multi-target balance and the like still exist. Disclosure of Invention In order to solve the problems, the invention provides a robot path planning method and a system based on target attention points, the invention adopts a global-level information-driven sampling decision and path planning strategy, candidate sampling points are generated by clustering the boundary grids of the known area and the unknown area, and attention target points with higher information value are screened by using an uncertainty gain index. In the movement process towards the attention target point, the robot acquires observation data in real time and updates a Gaussian process regression model, so that synchronous reconstruction of an environment field is realized, autonomous sampling of the robot can be effectively guided, information acquisition is maximized while a travel path is optimized, and the accuracy and efficiency of reconstruction of the environment field are remarkably improved. According to some embodiments, the present invention employs the following technical solutions: a robot path planning method based on target points of attention, comprising the steps of: according to laser data acquired by the robot, modeling an environmental field based on a Gaussian process; generating target attention candidate points at known-unknown boundaries of the environmental field based on a leading edge detection and clustering algorithm, and taking prediction variances at the candidate points as measures of field uncertainty; constructing a utility function integrating the map information gain, the field information gain and the path cost, and selecting a target attention point from candidate points by maximizing the utility function; An optimal path is generated using a TEB path planner. In an alternative embodiment, the process of modeling the environmental field based on a Gaussian process based on laser data acquired by a robot includes selecting a kernel functionFor any unobserved positionThe posterior predictive distribution of the environmental variables is: Wherein, the To train the covariance matrix between the samples,To train the covariance vector between the samples and the predicted points,Is the covariance matrix between the training points,The prediction quantity is driven by observation data provided by an environmental field sensor group; posterior mean value at each point in space based on environment variable An environment field reconstruction map of the whole area can be generated based on the posterior varianceAnd quantifying uncertainty of the reconstruction result. As an alternative embodiment, the process of generating target attention candidate points at known-unknown boundaries of the environmental field based on the leading edge detection and clustering algorithm comprises: acquiring all front points in the current occupied grid map through front edge detection, wherein the front edge points comprise boundary grids of explored areas and unknown areas, clustering the front edge points by using a clustering algorithm to obtain centroids of a plurality of front edge candidate areas, and forming a candidate point set by all centroids 。 Alternatively, the process of using the predicted variance at the candidate points as a measure of field uncertainty includes defini