Predictive distant operation and virtual control of computer numerical control machines

With the advancements of automation technologies, factory operations have shifted from labor-based to semi-automatic and fully automatic and may even become unmanned in the future. Therefore, the conditions of a factory can be monitored from a distance and the machines can be remotely controlled. This study investigated the predictive distant operation of a computer numerical control (CNC) machine virtually controlled with hand gestures. Similar previous studies were conducted only involving robotic arms. Nevertheless, the attempt in this study is a crucial step toward entirely Internet- or cloud-based manufacturing. In the proposed methodology, meaningful hand gestures corresponding to eight commonly used CNC operations were established. Kinect was then used to track the skeletal information of an operator, and based on this information, the joint angles of the operator were derived using the space vector method. Four rules were then proposed to predict the hand gesture of the operator according to the changes of the joint angles when making the hand gestures to control the CNC machine. An experimental system was established to demonstrate the applicability of the proposed methodology. According to the experimental results, the eight hand gestures were recognized with favorable accuracy. In addition, analysis of variance was performed and the results indicated that the performance of the proposed methodology is robust to the operators' differences.