Implementing an Automated Reasoning System for Multi-Robot Cooperation

We present an implementation of an automated reasoning system for multi-robot cooperation. A multi-robot cooperation environment is expressed by a set of multi-robot knowledge and time modal logic formulas. The reasoning procedure of our system is based on a so-called semantic method, that is, a set of multi-robot knowledge and time modal logic formulas is first translated, according to the possible world semantics, into the set of corresponding first-order clauses. Then, instead of checking the satisfiability of the given multi-robot knowledge and time formulas, we check the satisfiability of the set of translated first-order clauses by a general purpose first-order theorem proof procedure ME (the model elimination), augmented with the capabilities of handling transitive axioms and inequality predicates introduced by the translation. We also consider how to use the framework of the possible-world semantics to capture the notions of common knowledge and implicit knowledge, which are essential when considering reasoning, planning and cooperating problem solving in distributed and dynamically changing environments. We then discuss how to translate common knowledge and implicit knowledge into their corresponding first-order formulas. We apply the idea of theory resolution for reasoning about transitive axioms efficiently. We also show some experimental results of our automated reasoning system.