Manipulator path planning in 3-dimensional space

摘要

Present paper is aimed to work out efficient algorithm of multi-chain manipulator path planning in 3D space with static polygonal obstacles. The resulting solution is based on navigational maps approach. Using this approach, manipulator features are considered as intellectual agent, and reachability information is stored in compact form. This enables fast adaptation to arbitrary parameters of manipulator and workspace. The paper describes two algorithms: (ⅰ) a local walkthrough with obstacle avoidance, and (ⅱ) incremental navigational map building, performed at running stage. Both algorithms take an extensive use of the specific features of the task. Working simultaneously, they allow real-time manipulator path planning, as well as self-learning in idle mode. Algorithms are implemented as a demonstration program.