A Buy-Time Fault Tolerant Method in Distributed Multi-robot Systems

In this paper, a new buy-time fault tolerant method is proposed based on a bounding parameter estimation algorithm. The system parameters are confined within given bounds. The size of the bounds shrinks as more information becomes available. In this case, the reconfigurable control scheme can rely on the estimated parameter bounds to provide preliminary control to stabilize the system and slow down the deterioration of the system performance. Thus longer time can be gained for further identification and control reconfiguration to recover the system performance gradually. The experimental results based on a distributed multi-robot system prove that with this method, the safety critical system can buy some time before the final crash happens.