Hybrid vision-force guided fault tolerant robotic assembly for electric connectors

Abstract

To detect and recover the errors occurred in the task of mating connectors is vital in robotic wiring harness systems. In our previous work, a set-membership approach for static piecewise affine (PWA) system was proposed. Although using the static force model, errors can be detected effectively during mating connectors, there are still some mistaken detections and unrecognized faults, due to the insufficient sensor information and limitation of model. Before the mating task, there are various kinds of grasping errors. Only using force sensor is unable to detect the grasping errors. In this study, a new hybrid vision-force guided fault tolerant approach is proposed to improve the rate of error detection. More features from the vision system are chosen as the parameters of the fault tolerant system. Multiple sensors including a force sensor, encoders and an industrial vision system are assumed to acquire the necessary information of the method. The effectiveness of these methods is finally confirmed through experiments.