Testing the Cooperation of Autonomous Robotic Agents

Abstract

This article proposes an approach to testing the cooperative behaviour of autonomous software-based agents with safety-relevant tasks. It includes the definition of different model-based testing criteria based on the coverage of Coloured Petri Net entities as well as the automatic generation of appropriate test cases. The multi-objective optimization problem considered addresses both the maximization of interaction coverage and the minimization of the amount of test cases required. The approach developed for its solution makes use of genetic algorithms. The resulting automatic test case generation process is presented in this article together with the experiences gained by applying it to cooperating autonomous forklifts.