Leader-follower tracking for a network of gliding robotic fish using dynamic feedback linearization

Gliding robotic fish combine the strengths of both underwater gliders and robotic fish, resulting in long duration of operation and high maneuverability. In this paper we study the distributed tracking of a leader´s gliding path, represented by its velocity magnitude and gliding angle, by a multi-agent network when the leader´s state is available to only a subset of the followers. The high nonlinearities in the robot´s dynamics along with the coupling between its input channels and the gliding path parameters impose significant challenges in controlling the robot. Feedback linearization is one of the popular techniques when dealing with such systems. However, when considering the velocity magnitude and gliding angle of the robot as output, it is not possible to linearize the system using a static feedback control law. Therefore, dynamic feedback linearization is utilized for extending the state space of the robot, which enables the design of a distributed linear controller to solve the leader-follower tracking problem. Distributed estimators are designed for each agent to estimate the synthetic input of the leader and its linearized state. The virtual control for each agent is then mapped back to the actual input in order to track the leader´s velocity and gliding angle. The effectiveness of the proposed approach is verified with simulation results.