Extensions of motion planning algorithms
Embargo Date
2022-05-07
OA Version
Citation
Abstract
Sample-based motion planning algorithms can be applied to a broad range of circumstances in motion planning of robotics. Though sample-based algorithms are able to generate collision-free paths without the information of obstacles, they still have two weaknesses: one is that it is challenging to pass through narrow passages, which might result in path generation failures; the other is that a fixed search scope might lead to a waste of computational resource, which would result in low efficiency. In order to limit the search scope and improve the efficiency of paths generation of narrow passages, obstacles in the configuration space can be used to constrain the sampling scope and guide sampling. This thesis develops Obstacle Activation to identify polygonal obstacles that can be used to limit the search scope and Obstacle Exploration to obtain free points in a non-polygonal configuration space. These two methods are combined with several improved sampling-based algorithms to achieve the acceleration of the shortest path and feasible path generation. Finally, through a large number of simulations, it is verified that Obstacle Activation and Obstacle Exploration can effectively improve the speed of sampling-based algorithms significantly on both challenging scenarios with narrow passages and general cases.
Description
License
Attribution 4.0 International