Vision-force guided monitoring for mating connectors in wiring harness assembly systems

Correctly mating connectors is vital in robotic harness wiring systems. Although a static force model has been established based on force and position information during mating, there are difficulties to be overcome to implement a mating process monitoring task with insufficient training samples. It is difficult to obtain consistent features from successful mating samples under different conditions using existing models, which makes it difficult to recognize whether the mating process is correctly completed by learning few successful training samples. In this study, a simple new model is proposed to describe the connector mating process. More robust features are chosen as model parameters. It turns out that the model is more appropriate for the monitoring task, despite varied assembly conditions that affect the mating process. Multiple sensors, including a force sensor, encoders and two CCD cameras are used to acquire the necessary information for the model. A real-time algorithm based on a set-membership approach is used to implement the monitoring task. The effectiveness of these methods is confirmed through experiments. This study provides a simple but effective monitoring approach for a realtime industrial system integration application.