Design of intelligent mechatronical systems with high-level Petri nets

摘要

We present an approach for the integration of reinforcement learning methods into Petri net based specifications of robot behaviors. Our work aims at opening an existing design methodology of embedded systems for the design of autonomous mechatronical systems with adaptive behavior. In order to combine Petri nets and learning methods, we modeled Q-learning - a variant of reinforcement learning - with high-level Petri nets. The result can be integrated into Petri net models of autonomous mechatronical systems, e.g. behavior-based robots. For an evaluation of our approach, we have implemented a realistic application example, a part of the well-known robot contest 'capture the flag'. The example has been evaluated by simulation as well as on a physical system.