The use of robotic systems in remote and/or hazardous environments requires fundamental advances in software to guarantee that such systems will perform safely and reliably, even when component failures are inevitable. The goal of this work is to design intelligent motion coordination software that instills robots with the capability to use information about component failures to assess the impact of these failures on their motion capabilities, compare their current abilities with those required to perform an assigned task, and to implement appropriate behaviors. The design of such failure tolerant robotic systems relies on the proper incorporation of redundancy, and its intelligent control. The criteria of operation include: (a) failures must be anticipated, (b) performance must gracefully degrade in response to component failures and (c) behavior must be failsafe in the presence of catastrophic failures. This project will develop software that is capable of: (1) evaluating the degree of fault tolerance of a robotic system design, (2) improving the fault tolerance of existing robotic systems or determining optimally fault tolerant designs, and (3) operating robotic systems in a fault tolerant manner in anticipation of potential failures and seamlessly transitioning to optimal post-failure control.
