Prediction of assembly errors by three-dimensional geometric trajectory tracking

When an assembly error occurs in a robotic assembly operation, the assembly system is suspended during recovery from the error and short interruptions, termed as minor stoppage, takes place. Human operators are indispensable for quick recovery which could be a major obstacle in the aim toward unmanned assembly systems. An assembly error can also cause system failure. Even if recovery actions are automated, constant posttreatment of assembly errors is still required; therefore, when the improvement of assembly efficiency is strictly considered, the increase of cycle time becomes a serious problem. In this study, we propose a method for predicting whether or not the concerned assembly operation is successful immediately before a mating operation is initiated in the assembly system. Three-dimensional information about the position and orientation of a mated part and a base part is obtained during transport with two high-speed CCD (charge coupled device) cameras. Then, assembly errors are predicted by conducting pattern information processing of the data obtained from the geometric trajectory and their attributes. The effectiveness of the proposed approach is demonstrated by carrying out assembly tasks using a SCARA (selective compliance assembly robot arm)-type robot.