Mission Reliability Estimation for Repairable Robot Teams

Many of the most promising applications for mobile robots require very high reliability. The current generation of mobile robots is, for the most part, highly unreliable. The few mobile robots that currently demonstrate high reliability achieve this reliability at a high financial cost. In order for mobile robots to be more widely used, it will be necessary to find ways to provide high mission reliability at lower cost. Comparing alternative design paradigms in a principled way requires methods for comparing the reliability of different robot and robot team configurations. In this paper, we present the first principled quantitative method for performing mission reliability estimation for mobile robot teams. We also apply this method to an example robot mission, examining the cost-reliability tradeoffs among different team configurations. Using conservative estimates of the cost-reliability relationship, our results show that it is possible to significantly reduce the cost of a robotic mission by using cheaper, lower-reliability components and providing spares.