Fast Convergence RRT for Asymptotically-optimal Motion Planning

Recently, the optimal motion planning problem has attracted a considerable amount of attention, giving rise to new algorithms like RRG, RRT* and PRM*. However, these algorithms have some difficulty in handling the high-dimensional situation like manipulation, which needs a large amount of samples to explore a huge configuration space. In this context, we present a novel incremental sampling-based motion planning algorithm called Fast Convergence Rapidly-exploring Random Tree (FCRRT). Besides the guarantee to asymptotic optimality, our method has two key improvements: (1) the exploration and the optimization procedures are implemented and executed independently to retain the exploration strength of RRT that rapidly grows a random tree toward unexplored regions of the C-space, and (2), the Lazy-RRG technique is used to accelerate the convergence rate of the method. Experimental results indicate that FCRRT significantly improves the exploration rate and success rate, and finds paths of a similar quality much more quickly compared to RRT*.