An RRT planner is given 5 seconds to plan a path for a 6-DOF robot arm. After 1 second, it has found a feasible path from start to goal. What should the planner do with the remaining 4 seconds?
AStop and return the path immediately, because a feasible solution has been found
BContinue sampling and tree expansion to find shorter and smoother paths, with anytime optimization
CRestart with a different random seed to explore alternative paths
DSwitch to a deterministic planner like A* to refine the RRT solution
This distinction drives algorithmic choice in practice. Mobile robots in static environments favor PRM (build roadmap once per environment, then answer rapid goal queries). Robotic manipulators in semi-dynamic environments favor RRT (environment changes, rebuild on each plan request). Hybrid approaches use PRM for the mobile base and RRT for the arm, leveraging each algorithm's strength. The choice trades off single-query planning time against multi-query amortized cost.