Designing optimal trajectory planners for robotic communications

There are an increasing number of applications in which a robot must perform some task (e.g., autonomous mobile surveillance) and then transmit data back to a base station (e.g., video or sensor information). However the robot will often experience severe small-scale RF fading and so it must also save energy (e.g., mechanical and RF) while searching for the best transmit location. Our approach is to visualise the (flat-fading) spatial wireless channel as the realization of a random scalar field that assigns to each Cartesian coordinate vector q a channel gain H_q = |h_q|. Here h_q (complex and scalar) is the wireless channel from the robot (at coordinate q) to the base-station. In this article we develop two intelligent trajectory planners which attempt to move the robot to the closest location which has a sufficiently high channel gain. The first trajectory planner (called the Gradient Trajectory Planner) creates the trajectory following the gradient of the channel gain scalar field. The second trajectory planner (called the Potential Field Trajectory Planner) uses artificial potential functions to find locations inside a finite dimensional grid which have a high probability of having a sufficiently high channel gain.