Improved Coverage with Minimal Sensing

Presentation Abstract:

The classical problem of robot coverage is to plan a path that brings a given point on the robot within some fixed radius of every point in the workspace. We are interested in a variant of this problem in which we are only required to cover a portion of the workspace along its boundary but must deal with significant uncertainty in both sensing and actuation. In particular, we consider a differential-drive robot with access only to wheel encoders and to boundary sensors that trigger when entering or leaving the workspace. We capture uncertainty by assuming drift in the wheel sizes and wheel base of this robot. We show that it is possible to estimate these parameters based on the difference in time between predicted and observed boundary crossings, but only for a good choice of intermediate trajectory. We apply a linear analysis of observability to design a sequence of these trajectories that allow both calibration and coverage near the boundary. Finally, we validate our approach using experiments both in simulation and with a real robot.