Tree-Search-Based Any-Time Time-Optimal Path-Constrained Trajectory Planning With Inadmissible Island Constraints

Time-optimal path-constrained trajectory planning is of significant importance for enhancing the work efficiency of robotic systems. In particular, when inadmissible island constraints are considered, existing approaches are typically offline. In order to solve the problem online, this paper proposes a heuristic tree-search-based any-time time-optimal trajectory planning algorithm to achieve incremental computation of feasible trajectories. In a limited planning period, the proposed algorithm iteratively generates feasible trajectories, as optimal as possible, until the period terminates. This any-time performance of the proposed algorithm ensures that feasible trajectories can be obtained in real time and even time-optimal trajectories can be obtained if the planning period is long enough for searching. Experimental results on active-casters-based omnidirectional wheeled mobile robots demonstrate the validity of the proposed algorithm.