Analysis of robots for hazardous environments

Reliability analysis of fault tolerant systems often ignores the small probability that a failure might not be detected or, if detected, may not be properly handled. The probability that the failure is detected and properly handled is called coverage. Inclusion of coverage in reliability analysis is especially important when analyzing critical systems, systems which for some reason are not easily reparable, or systems whose failure can result in serious damage to the system or its surroundings. One example of a system which can cause such damage is a robot manipulator arm. Robots are being increasingly employed in remote and hazardous environments such as in space and in nuclear waste cleanup, and can exhibit a wild response to subsystem failure, damaging themselves and/or their surroundings. Addition of redundancy to such systems can increase their reliability by allowing continued operation in the presence of faults (provided that the fault is covered), an advantage in a system where repair is difficult or impossible. Coverage models have been used to analyze the behavior of fault-tolerant computer systems in the presence of faults, providing an estimate of the relative probability of an uncovered vs. a covered component failure (given that a fault has occurred). This paper extends the use of coverage models to the basic components of the joint of a robot and presents data utilizing the calculated coverage for a three-joint robot manipulator arm designed to operate in the plane.