Sampling-Based Motion Planning

The field studies how robots can find collision-free paths and dynamically feasible motions when their configuration spaces are too large or complex to search exhaustively. Common methods include tree-based planners such as RRT and RRT*, roadmap methods such as PRM, and kinodynamic variants that account for motion constraints; informed and learned samplers guide search toward promising regions. Work also extends these planners to manipulation, multi-robot coordination, moving obstacles, and uncertain or partially observed environments, often using parallel computing to speed up exploration and collision checking. A central tension is balancing fast, reliable solutions with path quality and safety guarantees, especially in high-dimensional or tightly constrained settings.