Abstract: An expanded similarity space (ESS) method for case-based reasoning and fault recovery is proposed by combining LID and CMM. It is the key to finding error nodes according to fault symptoms for the ESS method. The ESS with the error nodes is derived, and then the errors can be tested in order of the probabilities of error nodes in the ESS. This method need not follow the order of the depth of nodes to test error nodes, which reveals the phenomenon that some errors are more common in nature. The reasoning model and algorithms for the ESS method are presented. Simulation result shows the average times of testing errors for the ESS method are smaller than those of LEAF. It means the ESS method can shorten the time of the multiple robots system for fault recovery, so the ESS method has a better capability.

Key words: expanded similarity space method, case-based reasoning, fault recovery, error node, multiple robots system