MOTION PLANNING IN DYNAMIC ENVIRONMENTS USING VELOCITY OBSTACLES

This paper presents a method for robot motion planning in dynamic envi ronments. It consists of selecting avoidance maneuvers to avoid static and moving obstacles in the velocity space, based on the current posi tions and velocities of the robot and obstacles. It is a first-order m ethod, since it does not integrate velocities to yield positions as fu nctions of time. The avoidance maneuvers are generated by selecting ro bot velocities outside of the velocity obstacles, which represent the set of robot velocities that would result in a collision with a given obstacle that moves at a given velocity, at some future time. To ensur e that the avoidance maneuver is dynamically feasible, the set of avoi dance velocities is intersected with the set of admissible velocities, defined by the robot's acceleration constraints. Computing new avoida nce maneuvers at regular time intervals accounts for general obstacle trajectories. The trajectory from start to goal is computed by searchi ng a tree of feasible avoidance maneuvers, computed at discrete time i ntervals. An exhaustive search of the tree yields near-optimal traject ories that either minimize distance or motion time. A heuristic search of the tree is applicable to on-line planning. The method is demonstr ated for point and disk robots among static and moving obstacles, and for an automated vehicle in an intelligent vehicle highway system scen ario.