Vision-Based Output-Feedback Leader-Follower Rigid Formation Tracking Control for Multiple Mobile Robots

This article presents a novel vision-based rigid formation framework for multiple mobile robot systems. The proposed framework integrates multiple formation tracking models, facilitating the achievement of rigid formation while addressing field-of-view (FOV) constraints. To resolve the tradeoff issue between visibility maintenance and rigid formation when the leader's trajectory exhibits large curvature, we propose a continuous switching strategy. This strategy prioritizes ensuring leader visibility, then restores the rigid formation geometry after traversing the large curvature trajectory segment. Considering the challenges posed by FOV constraints and unknown nonlinear dynamics, we introduce an output-feedback control scheme. This scheme involves a fixed-time model-free observer that estimates the essential high-order error dynamics and the leader's velocities using only data from onboard cameras. Building upon these estimations, a novel output-feedback control protocol is developed that integrates homeomorphism mapping and time-delay estimation techniques. Moreover, an adaptive technique is employed to further enhance control performance by dynamically updating control gains. The proposed framework provides a comprehensive solution to the limitations of existing vision-based leader-follower methods in achieving rigid formation with limited sensing capability. Finally, real-world experiments have been carried out to validate the effectiveness of the proposed framework.