Leader–Follower Navigation While Preserving Network Connectivity Without Explicit Communication in the Presence of Moving Obstacles

Abstract

This paper proposes a leader–follower navigation method for a group of robots in the presence of moving obstacles without explicit communication over wireless networks. The proposed method consists of a control network and a sensing network. When moving obstacles are predicted to pass nearby, robots reconfigure the control network by deactivating the network links in the predicted area in a decentralized manner while maintaining network connectivity. For control network links whose deactivation may cause a split of the group (i.e., the loss of connectivity of the control network), the robots are controlled to remain within each other’s sensing range during and after deactivation until the obstacle has passed, and then execute merging maneuvers to reconnect the subgroups. Although a temporary loss of connectivity in the sensing network during the obstacle’s passage is inevitable, robots on either side of the obstacle try to stay within each other’s sensing range using pre-obstruction sensor data to facilitate reconnection once the line of sight is restored. We perform extensive simulations to evaluate the proposed method’s performance in terms of success rates, with and without split–merge behaviors, and the time required to navigate obstacle-rich environments. Additionally, experiments using real quadrotors validate the method’s feasibility.