Analyzing unsynthesizable specifications for high-level robot behavior using LTLMoP

Abstract

Recent work in robotics has applied formal verification tools to automatically generate correct-by-construction controllers for autonomous robots. However, when it is not possible to create such a controller, these approaches do not provide the user with feedback on the source of failure, making the experience of debugging a specification somewhat ad hoc and unstructured, and a source of frustration for the user. This paper describes an extension to the LTLMoP toolkit for robot mission planning that encloses the control-generation process in a layer of automated reasoning to identify the cause of failure, and targets the user's attention to flawed portions of the specification. © 2011 Springer-Verlag.