Realization of a Cooperative Human-Robot-Picking by a Learning Multi-Robot-System Using BDI-Agents

This article presents a process model for a picking environment which allows robots and humans to collaborate while simultaneously allowing the robots to increase their performance through directed integration of human feedback. Our approach is an adaptive multi-robot-system based on Active Component Shells to handle the automatic detection and gripping of objects. The learning and picking processes are guided by a warehouse management system and are supported by a computation cluster. The system integrates human pickers into these processes to support robotic learning and to ensure process stability as a fallback. The picking robots and the IT systems are realized by a BDI multiagent-system communicating via MQTT