Randomized Kinodynamic Planning

LaValle & Kuffner [] present an application of a randomized technique to the problem of kinodynamic planning. Their algorithm constructs Rapidly-exploring Random Trees (RRTs) in a high dimensional state space that encompasses both first order constraints resulting from the physically-based system dynamics as well as global kinematic constraints due to obstacles in the environment. By coupling and simultaneously addressing the problems of finding a basic path in configuration space and finding a controller satisfying the system dynamics while exploiting the benefits of randomized techniques, trajectories for a broad range of systems operating in cluttered environments can be computed in an efficient manner.