Distributed Model Predictive Control for Multiple Unmanned Ground Vehicles Formation with Packet Loss

Combining distributed model predictive control (DMPC) with the terminal invariant set, this paper designs a distributed control scheme for multiple unmanned ground vehicles (UGVs) formation. When unexpected situation such as road construction occurs, UGVs can change formation to cope with emergency. It is an important issue to accurately estimate the lost data and continue to complete the task when data packet loss occurs in the process of UGVs communication. In order to ensure stability of the controller, the terminal invariant set and the local feedback control law are added at the end of predictive time domain, and then solved by using linear matrix inequalities (LMI). Considering the formation task, collision avoidance constraint, control input constraint and terminal constraint, an optimization model is formed. By solving the optimization model, control input and optimal estimation of future path can be confirmed. Then, the updated path is sent to the adjacent UGVs for collision avoidance. Finally, combining the time window, data estimation based on local feedback control law is discussed when the data packet loss occurs.