A Real-Time Knowledge Scheme for Sensory-Controlled Robot Assembly Tasks

In the field of assembly automation with industrial robots the research on the application of artificial intelligence techniques is getting increasing importance. This paper describes the hierarchical structure of a knowledge—based sensory-controlled robot assembly system under development which is capable to plan and execute assembly tasks under real—time requirements. The hybrid knowledge representation scheme combining the rule—based and object—oriented approach to represent the assembly domain—specific knowledge is discussed. Furtheron, the knowledge processing concept based upon the representation scheme is explained. A first prototype of the system has been implemented in a real robotic test—bed. Several peg/hole part mating sequences validated the capability of the system to execute assembly tasks in an uncertain environment using sensory information.