Multi-robot behaviors with bearing-only sensors and scale-free coordinates

Abstract

This thesis presents a low-cost multi-robot system for large populations of robots, a new coordinate system for the robot based on angles between robots and a series of experiments validating robot performance. The new robot platform, the r-one will serve as an educational, outreach and research platform for robotics. I consider the robot's bearing-only sensor model, where each robot is capable of measuring the bearing, but not the distance, to each of its neighbors. This work also includes behaviors demonstrating the efficiency of this approach with this bearing-only sensor model. The new local coordinate systems based on angular information is introduced as scale-free coordinate system . Each robot produces its own local scale-free coordinates to determine the relative positions of its neighbors up to an unknown scaling factor. The computation of scale-free coordinates is analyzed with hardware and simulation validation. For hardware, the scale-free algorithm is tailored to low-cost systems with limited communication bandwidth and sensor resolution. The algorithm also uses a noise sensitivity model to reduce the impact of noise on the computed scale-free coordinates. I validate the algorithm with static and dynamic motion experiments.