A fast collision-free path planning method for a general robot manipulator

We propose a practical path planning method that can find collision-free paths quickly for a general robot manipulator. The proposed method repeats the global search and serial local search. The global search tries to find a sequence of intermediate goals (sub-goals). The serial local search tries to sequentially find local paths between each two adjacent sub-goals. When serial local search fails to find one of local paths, the method backtracks to the global search. The main feature is that the search range for each local search is restricted and is gradually enlarged at re-trials. We conducted simulations in 3D C-space. The result shows that proposed method can find paths 200 times faster than traditional discrete C-space method.