Monte-Carlo Robot Path Planning

Path planning is a crucial algorithmic approach for designing robot behaviors. Sampling-based approaches, like rapidly exploring random trees (RRTs) or probabilistic roadmaps, are prominent solutions path problems. Despite its exponential convergence rate, RRT can only find suboptimal paths. On the other hand, $\text {RRT}^*$ , widely-used extension to RRT, guarantees completeness finding optimal paths but suffers in practice from slow complex environments. Furthermore, real-world robotic environments often partially observable with poorly described dynamics, casting application of tasks suboptimal. This letter studies novel formulation popular Monte-Carlo tree search (MCTS) algorithm planning. Notably, we study Planning (MCPP) by analyzing and proving, on one part, rate fully Markov decision processes (MDPs), feasible MDPs (POMDPs) assuming limited distance observability (proof sketch). Our contribution allows us employ recently proposed variants MCTS different exploration strategies experimental evaluations simulated 2D 3D 7 degrees freedom (DOF) manipulator, as well task, demonstrate superiority MCPP POMDP tasks.