Minimize Collision Risk Global Planner Based Voronoi Graph for Mobile Robot

Huang Zhen,Wen Wang,Xu Jinpeng,Liu Zhihui
DOI: https://doi.org/10.1109/ICRCV55858.2022.9953252
2022-09-25
Abstract:Ensure safe, reliable, fast and stable robot navigation which requires stable and efficient motion planning algorithm support. In order to reduce the collision risk during robot navigation, we propose minimize collision risk global planner for mobile robot, based on voronoi graph to extract Tyson polygon vertices. Using multiple vertices to form a planning search graph, the collision risk of each edge is calculated i.e. the integral of the distance to the obstacle is used as the weighted value of the edge, and the search is performed using A* to obtain the planning path with the lowest collision risk globally at the current moment. In order to make the planned path more consistent with robot kinematics, we designed the collision and smooth cost function to optimize the path using gradient descent, and the searched path is passed to the local planner after path smooth. The test results on the actual robot show that the robot can effectively avoid narrow areas and effectively reduce the collision rate and replanning times in dynamic environments.
Computer Science,Engineering
What problem does this paper attempt to address?