Skill-based multi-agent control for safe and effective human-robot collaboration

Abstract Due to their complexity, various tasks in production technology are still carried out manually. Support by robotic systems often fails due to the high safety requirements and the associated integration costs. In this paper, a novel agent-based approach for the design and integration of human-robot teams is presented, which models the autonomous behavior and task sharing between humans and robots symmetrically and structures it on several levels. An essential component here is the uniform description of human and robotic context-specific autonomous skills and the software modules realized with them, which can be freely configured for the respective task, taking into account the respective safety conditions. The reusability of interoperable software modules saves costs and ensures interchangeability between different manufacturers and systems. The approach is implemented using the example of a collaborative assembly scenario and demonstrated for the assembly of a toy truck in the SmartFactory Kaiserslautern.