The use of mobile robots has increased over the last years in a wide range of applications, which go from industrial work to more domestic tasks, such as vacuum cleaning or grass cutting. Due to the benefits that these systems bring and the complexity of the environments where they are used, it is necessary to improve their reliability, which can be done by adding fault detection systems that are capable of diagnosing the operating state of the robot and react according to it. This thesis centers on the problem of failures on cooperative mobile robots, by designing a new low-cost architecture of fault detection and diagnosis, which is later implemented. This system allows both the control structure and the operators of the robot, to take into account the actual operating state of the mobile robot, by giving more information to them. The architecture is first evaluated using a simulation platform in MATLAB, which also helps to analyze the performance of the system. Then, the fault detection and diagnosis system is implemented on a cooperative group of small and resource-limited mobile robots, showing an excellent performance in the identification of the different faults.