Fault-Tolerant Multi-Robot Target Pursuit with Fault Detection and Isolation Based on Mobile Adaptive Networks

Multi-robot target pursuit is a typical problem in multi-robot systems and can be solved by mobile adaptive networks in distributed way. For the importance and difficulty in improving fault tolerance of multi-robot systems, many adaptations of mobile adaptive networks have emerged. This paper introduces fault detection and isolation to the standard (with the weighted average aggregation rule) and three existed adaptations (with the coordinate-wise median, geometric median, and centerpoint aggregation rules) of mobile adaptive networks, which uses the differences of the theoretical estimated values and the actual estimated values to predict normal neighbor set in mobile adaptive networks, and carries out target pursuit. Fault-tolerant multi-robot target pursuit algorithms with fault detection and isolation based on mobile adaptive networks are proposed to remove the interference of the faults in the networks to achieve better performance in target pursuit. Simulation experiments have been carried out with different numbers of faulty robots in the original algorithms and our improved algorithms. Our proposed algorithms are proved to perform better in convergence and accuracy than the original algorithms, especially when the number of faulty robots increases, and the improvement effects on target pursuit algorithms with different aggregation rules can be different.