Physics-Based Output-Feedback Consensus-Formation Control of Networked Autonomous Vehicles

We describe a control method for multi-agent vehicles, to make them adopt a formation around a non-pre-specified point on the plane, and with common but non-pre-imposed orientation. The problem may be considered as part of a more complex maneuver in which the robots are summoned to a rendezvousRendezvous to advance in formation. The novelty and most appealing feature of our control method is that it is physics-based; it relies on the design of distributed dynamic controllers that may be assimilated to second-order mechanical systemsMechanical systems. The consensusConsensus task is achieved by making the controllers, not the vehicles themselves directly, achieve consensusConsensus. Then, the vehicles are steered into a formation by virtue of fictitious mechanical couplings with their respective controllers. We cover different settings of increasing technical difficulty, from consensusConsensus formation controlFormation control of second-order integrators to second-order nonholonomicNonholonomic vehicles and in scenarii including both state- and output-feedback control. In addition, we address the realistic case in which the vehicles communicate over a common WiFi network that introduces time-varying delaysDelay. Remarkably, the same physics-based method applies to all the scenarii.