Heterogeneous robots: Model Predictive Control for bearing-only formation and tracking

Multi-agent systems are systems composed by more than one autonomous robots which usually work under the assumption that they can communicate sending and receiving positions of other robots that operate in the network. The introduction of this kind of systems is due to the fact that in many situations it is preferable to use more than one robot in order to reach more complex goal without the help of the humans, especially in dangerous situations. In this thesis, the focus is on the heterogeneous robots which are robots whose components are heterogeneous in terms of actuation capabilities, even if it is assumed they can receive bearing information with respect to the other agents in the network. Hence, it is developed an heterogeneous MAS composed by 2 UGVs and 2 UAVs. The goals of the thesis is that the formation has to be maintained and the four agents has also to track a desired trajectory through a leader follower approach based on bearing-only implemented using MPC controllers. The role of the leader is to track the desired trajectory while the followers have to form and maintain the formation also during the tracking. The followers do not know the trajectory to be tracked, nor the distance to the other agents and the leader. The approach is based on decentralized leader follower control with bearing-only. The controllers used are the Model Predictive ones since this type of control allow to prevent the critical situations, solving an online optimization problem at each time instant to select the best control action that drives the predicted output to the reference. The proposed approach is implemented in Matlab and Simulink and the results obtained by the simulations will be discussed.