Machine Vision and Six-Dimensional Force Recognition for Assembly Hole-Finding Strategies

In response to the current problems of bolt assembly in industrial automation, which can easily lead to jamming and hole-finding failure, the hole-finding method for bolt assembly is investigated. According to the different contact positions of the bolts, the force and torque values generated are different, and a method is proposed to determine the position of the bolts by identifying the magnitude and direction of the six-dimensional force. A monocular camera is used to obtain information on the position of the assembly hole to complete the initial positioning, a BP neural network is used to judge the assembly state, bolt position information is obtained from the feedback of force information, and a Gaussian fit is used to obtain the bolt movement distance to adjust the robot end motion. The results show that the robot can be reliably guided to complete the bolt assembly hole-finding within 30 seconds by visual guidance and feedback of force information.