Three-Layered Control Architecture for Microassembly with Human-Robot Task Plan Interaction

Assembly manipulation in the micro world presents challenges not found in the macro domain. A three-layered control architecture for microassembly with human-robot task plan interaction is proposed in this paper. The architecture combines the abilities of human decision making and task planning with visual servo strategies to ensure the reliability and automaticity of micromanipulation. It has three control levels: the behaviour layer interacts with the real physical world to control actuators and sensors. The task layer specifies assembly goals by human task planning. As the middle level, the strategy layer yields visual servo schemes from abstract assembly goals and expands them into low-level behaviour commands. TaskML, the task markup language, is designed based on XML for human task planning. Task plan tree (TPT) data specifies microassembly tasks underlying TaskML. Microassembly task planner (MTP) software has been developed to support TaskML editing and task plan tree building