Wheeled mobile robots (WMRs) are of great importance. Therefore, it is necessary to make sure that they are not defected. But, in case of failures, the diagnosis task is very important to predict then solve the problem. The most useful techniques in diagnosis are observers which are based on the observability of the monitored system that is not usually ensured by WMR. Thus, to overcome this drawback, an intelligent cooperative diagnosis algorithm is proposed and tested for a group of mobile robots. The diagnosis algorithm is based on robust adaptive unknown input observer applied on unobservable robot. The local non-observability of each robot is solved by cooperative communication. The idea consists on considering all WMRs as a Large Scale System (LSS) even these robots may have not common task. Then, the LSS is decomposed into subsystems that everyone refers to each robot communicating with its neighbors. Next, a design of cooperative interconnected systems is studied to reassure the new condition of observability. Besides, Fast Adaptive Fault Estimation (FAFE) algorithm is proposed to improve the performances of the fault estimation. Finally, to illustrate the efficiency of the proposed algorithm, a model of three-wheel omnidirectional mobile robot is presented.
Read full abstract