CN-121977579-A - Unmanned vehicle approaching reconnaissance method and system based on dynamic shielding path planning

CN121977579ACN 121977579 ACN121977579 ACN 121977579ACN-121977579-A

Abstract

The invention relates to an unmanned vehicle approaching reconnaissance method and system based on dynamic shielding path planning, and belongs to the technical field of unmanned vehicle autonomous navigation and path planning. A method for detecting the approaching of an unmanned vehicle based on dynamic shielding path planning comprises the steps of identifying candidate shielding objects based on real-time acquisition of environmental data of the unmanned vehicle, screening effective shielding objects according to effectiveness judging conditions and generating position and shape information of the effective shielding objects, planning a detection path on an environmental cost map by combining the current position, target position and shielding object information of the unmanned vehicle, controlling the unmanned vehicle to travel along the path, monitoring the path based on the environmental data updated in real time, and re-planning the path and controlling the unmanned vehicle to travel when re-planning conditions are met. The problem of current unmanned vehicles lack in the reconnaissance process to the recognition precision low, difficult real-time planning hidden travel path leads to exposing risk high to the shelter in the environment is solved.

Inventors

ZHANG XIAOPING
JIANG CHENXING
SU BO
JI CHAO
LU PEIYUAN
WANG DONGJI
XIONG GAO
ZHAO JUNRUI
LV GUOWEI
Peng Binsen

Assignees

中兵智能创新研究院有限公司

Dates

Publication Date: 20260505
Application Date: 20260407

Claims (10)

1. An unmanned vehicle approaching reconnaissance method based on dynamic shielding path planning is characterized by comprising the following steps of: identifying candidate shields in the environment based on environmental data acquired in real time by the unmanned vehicle; screening effective shielding objects meeting shielding requirements from the candidate shielding objects according to preset effectiveness judging conditions; Generating mask position information and mask shape information according to the effective mask; Planning a reconnaissance path on a pre-constructed environment cost map according to the current position, the target position, the position information of the shielding object and the shape information of the shielding object of the unmanned vehicle; And controlling the unmanned vehicle to travel along the reconnaissance path, monitoring the reconnaissance path based on the environmental data updated in real time, and when the preset rescheduling condition is met, rescheduling the reconnaissance path and controlling the unmanned vehicle to travel along the rescheduled reconnaissance path.
2. The method of claim 1, wherein identifying candidate masks in the environment based on environmental data collected in real-time by the drone comprises: acquiring image data acquired by a vision sensor carried by an unmanned vehicle and point cloud data acquired by a laser radar; inputting the image data into a pre-trained deep learning recognition model, and extracting a visual recognition result, wherein the visual recognition result comprises a target area and category confidence thereof; performing cluster analysis on the point cloud data to obtain a plurality of point cloud clusters, and calculating geometric characteristic parameters of each point cloud cluster; And carrying out space matching on the target area and the point cloud cluster, carrying out consistency verification according to the category confidence and the geometric characteristic parameter, and determining the verified target area as the candidate shielding object.
3. The method of claim 2, wherein performing consistency verification based on the category confidence and the geometric feature parameter, determining a verified target area as the candidate mask comprises: calculating the spatial overlapping degree of each target area and each point cloud cluster, and establishing a matching relationship between the target areas with the spatial overlapping degree larger than a preset overlapping threshold and the point cloud clusters; Aiming at a target area and a point cloud cluster with a matching relationship, calculating a multi-mode consistency score based on the category confidence of the target area and the geometric characteristic parameters of the point cloud cluster; And when the multi-mode consistency score is larger than a preset score threshold value, judging that the consistency verification is passed, and determining the target area as a candidate shelter.
4. The method of claim 1, wherein generating shutter position information from the effective shutter comprises: Acquiring a point cloud cluster corresponding to each effective shielding object; Determining a coordinate reference value of the point cloud cluster on each space dimension according to the space coordinates of all points in the point cloud cluster; Determining the three-dimensional space mass center of the point cloud cluster based on the coordinate reference values in each space dimension; and taking the three-dimensional space centroid as the shade position information of the effective shade.
5. The method of claim 1, wherein generating mask shape information from the effective mask comprises: Acquiring a point cloud cluster corresponding to each effective shielding object; calculating a covariance matrix of the point cloud cluster in space based on the space coordinates of each point in the point cloud cluster; performing eigenvalue decomposition on the covariance matrix to obtain three eigenvalues and eigenvectors corresponding to each eigenvalue, wherein the three eigenvectors are used as three mutually perpendicular main directions, including a first main direction, a second main direction and a third main direction; projecting all points in the point cloud cluster to the first main direction, the second main direction and the third main direction respectively, and calculating the projection length in each main direction to obtain the length, the width and the height of the effective shielding object, wherein the length is the projection length in the first main direction, the width is the projection length in the second main direction and the height is the projection length in the third main direction; Taking the first main direction as the direction of the effective shelter; The length, width, height and orientation of the effective mask are taken as the shape information of the effective mask.
6. The method of claim 1, wherein planning a scout path on a pre-constructed environmental cost map comprises: Determining a path searching space on an environment cost map by taking the current position of the unmanned vehicle as a starting point and the target position as an end point; discretizing the path search space into a plurality of path nodes; Determining an exposed area in the path search space based on the mask position information and the mask shape information; calculating the exposure risk cost of each path node based on the distance from each path node to the nearest exposure area, the distance to the nearest effective shield and the path curvature between the path node and the previous path node; calculating a shield score of each path node based on the distance from each path node to the effective shield and the relative angle between the effective shield and the reconnaissance object; Scout paths are searched in a path search space at a plan in the path search space based on the exposure costs and the shield scores.
7. The method of claim 6, wherein the exposure cost for each path node is calculated as follows; ; Wherein, the As an exposure risk cost for path node n, For the distance of path node n to the nearest exposed area, For the distance of path node n to the nearest active mask, As a path curvature penalty term for path node n, 、、 Is a weight coefficient.
8. The method of claim 6 wherein calculating a mask score for each path node is represented by the formula; ; Wherein, the For the mask score of the path node n, For the type weight of the i-th effective mask, As a function of the distance decay from the path node n to the i-th effective mask, Is a function of the relative angular availability of the ith effective mask to the direction of the object under investigation.
9. The method of claim 1, wherein the preset rescheduling conditions comprise: a new obstacle is detected on the scout path or an active shelter on the scout path disappears.
10. An unmanned vehicle approach reconnaissance system based on dynamic masking path planning, the system comprising: the identification module is used for acquiring environment data in real time based on the unmanned vehicle and identifying candidate shields in the environment; The screening module is used for screening effective shielding objects meeting shielding requirements from the candidate shielding objects according to preset effectiveness judging conditions; a generating module for generating mask position information and mask shape information according to the effective mask; The planning module is used for planning a reconnaissance path on a pre-constructed environment cost map according to the current position, the target position, the position information of the shielding object and the shape information of the shielding object of the unmanned vehicle; And the control module is used for controlling the unmanned vehicle to travel along the reconnaissance path, monitoring the reconnaissance path based on the environmental data updated in real time, and when the preset rescheduling condition is met, rescheduling the reconnaissance path and controlling the unmanned vehicle to travel along the rescheduled reconnaissance path.

Description

Unmanned vehicle approaching reconnaissance method and system based on dynamic shielding path planning Technical Field The invention relates to the technical field of unmanned vehicle autonomous navigation and path planning, in particular to an unmanned vehicle approaching reconnaissance method and system based on dynamic shielding path planning. Background In an unmanned vehicle approaching reconnaissance task, concealment is a key element for ensuring task success and platform safety. The prior art mainly relies on a preset route or a simple obstacle avoidance algorithm to carry out path planning, and generally considers obstacles as negative factors to be avoided, but does not fully utilize naturally-occurring shields in the environment, such as rocks, bushes, buildings and the like, as shielding resources. In the existing method, a single sensor mode is adopted in the aspect of shielding object identification, the identification precision is limited, and three-dimensional geometric information and space orientation of the shielding object are difficult to accurately obtain, so that the actual shielding effectiveness of the shielding object cannot be effectively estimated through path planning. On the path optimization objective, the traditional method focuses on the shortest distance or shortest time, does not incorporate exposure risk quantification into the cost function, and does not build a dynamic correlation model between the type of shade, distance, azimuth angle and shield effectiveness. More importantly, the existing system lacks an online re-planning mechanism, and when the environment changes, the path cannot be adjusted in real time, so that the unmanned vehicle is exposed in the visual field range of the reconnaissance object in the running process, and the discovered probability is remarkably increased. Disclosure of Invention In view of the above analysis, the embodiment of the invention aims to provide an unmanned vehicle approaching reconnaissance method and system based on dynamic shielding path planning, which are used for solving the problems that the existing unmanned vehicle has low recognition precision on a shielding object in the environment in the approaching reconnaissance process, and the exposure risk is high because the most concealed travelling path is difficult to plan in real time. In one aspect, an embodiment of the present invention provides an unmanned vehicle approach reconnaissance method based on dynamic shielding path planning, including: Based on the unmanned vehicle, collecting environmental data in real time, and identifying candidate shields in the environment; screening effective shielding objects meeting shielding requirements from the candidate shielding objects according to preset effectiveness judging conditions; Generating mask position information and mask shape information according to the effective mask; Planning a reconnaissance path on a pre-constructed environment cost map according to the current position, the target position, the position information of the shielding object and the shape information of the shielding object of the unmanned vehicle; And controlling the unmanned vehicle to travel along the reconnaissance path, monitoring the reconnaissance path based on the environmental data updated in real time, and when the preset rescheduling condition is met, rescheduling the reconnaissance path and controlling the unmanned vehicle to travel along the rescheduled reconnaissance path. Further, based on the unmanned vehicle collecting environmental data in real time, identifying candidate masks in the environment includes: acquiring image data acquired by a vision sensor carried by an unmanned vehicle and point cloud data acquired by a laser radar; inputting the image data into a pre-trained deep learning recognition model, and extracting a visual recognition result, wherein the visual recognition result comprises a target area and category confidence thereof; performing cluster analysis on the point cloud data to obtain a plurality of point cloud clusters, and calculating geometric characteristic parameters of each point cloud cluster; And carrying out space matching on the target area and the point cloud cluster, carrying out consistency verification according to the category confidence and the geometric characteristic parameter, and determining the verified target area as the candidate shielding object. Further, performing consistency verification according to the category confidence and the geometric feature parameter, and determining the verified target area as the candidate mask includes: calculating the spatial overlapping degree of each target area and each point cloud cluster, and establishing a matching relationship between the target areas with the spatial overlapping degree larger than a preset overlapping threshold and the point cloud clusters; Aiming at a target area and a point cloud cluster with a matching relationship, calculatin