Abstract

Occlusion problems and environmental dependencies limit the use of motion capture systems or GPS sensors for global tracking in swarm robots. We address the demand for swarm operation by providing a robust and flexible on-board localization framework for multi-robot systems. Our method is based on a tag robot with one ultrawideband (UWB) sensor and an anchor robot with three UWB sensors. By using these sensors, the anchor robot may automatically establish the tag robot's position, saving the two robots from having to communicate directly in order to construct a localization framework. We use a dual Monte-Carlo localization (MCL) technique to accurately mimic the tag robot's nimble movements. We verify the effectiveness of our approach with real-world tests and simulations with a two-drone setup. The dual MCL method that has been suggested reliably outperforms both conventional particle filter and extended Kalman filter techniques in terms of producing highly accurate predictions over a wide range of tag robot speed profiles. Our system promotes flexibility and robustness in swarm operations and provides a workable method for multi-robot localization.