Motion planning and coordination of mobile robot behavior for medium scale distributed wireless network experiments

In this research, a computerized motion planning and control system for multiple robots is presented. Medium scale wheeled mobile robot couriers move wireless antennas within a semicontrolled environment. The systems described in this work are integrated as components within Mobile Emulab, a wireless research testbed. This testbed is publicly available to users remotely via the Internet. Experimenters use a computer interface to specify desired paths and configurations for multiple robots. The robot control and coordination system autonomously creates complex movements and behaviors from high level instructions. Multiple trajectory types may be created by Mobile Emulab. Baseline paths are comprised of line segments connecting waypoints, which require robots to stop and pivot between each segment. Filleted circular arcs between line segments allow constant motion trajectories. To avoid curvature discontinuities inherent in line-arc segmented paths, higher order continuous polynomial spirals and splines are constructed in place of the constant radius arcs. Polar form nonlinear state feedback controllers executing on a computer system connected to the robots over a wireless network accomplish posture stabilization, path following and trajectory tracking control. State feedback is provided by an overhead camera based visual localization system integrated into the testbed. Kinematic control is used to generate velocity commands sent to wheel velocity servo loop controllers built into the robots. Obstacle avoidance in Mobile Emulab is accomplished through visibility graph methods. The Virtualized Phase Portrait Method is presented as an alternative. A virtual velocity field overlay is created from workspace obstacle zone data. Global stability to a single equilibrium point, with local instability in proximity to obstacle regions is designed into this system.