Exposing Off-Nominal Behaviors in Multi-Robot Coordination

Abstract

Often software in robotics systems is susceptible to unexpected and unforeseen behaviors called off-nominal behaviors (ONBs) and these ONBs can affect the reliability or safety of the systems. While some work is done on exposing ONBs in a system, there has been little research conducted on exposing ONBs when multiple robots perform a task together. In this paper, we propose a combinatorial based approach to expose ONBs in such multi-robot coordination tasks during the requirements engineering phase. Our approach separates system level analysis and coordination level analysis, and generates combinations that need to be manually analyzed for ONBs. To evaluate the effectiveness of our approach, we conducted an empirical study with a set of requirements that have three coordination tasks. The results of our study show that our approach offers a means for ONB knowledge acquisition and reduces significant human effort and time required for exposing ONBs.