Distributed formation control with pose estimation in multi-robot systems

Abstract

Distributed formation control is motivated by many applications of multi-robot systems. Many existing control algorithms assume that each robot is equipped with sophisticated sensors that can measure robot's pose, i.e., position and orientation. In this paper, however, we consider that each robot has only ranging sensors (such as ultrasonic ones) and coarse positioning sensors (such as GPS). We apply continuous-time extended Kalman filter (CEKF) to estimate the robot's pose. By exploiting the collaboration with the neighboring robots, a virtual structure based distributed algorithm is adopted for achieving global formation. We evaluate the impact of neighborhood size on the control performance by extensive simulations. Moreover, we set up a platform and conduct experiments for different scenarios to verify the proposed algorithms.