Digital twin based cycle time prediction for robotic arm

This paper proposes a novel approach for accurate cycle time prediction in industrial robots by using a digital twin and a deep neural network (DNN). Conventional methods do not account for real-time conditions, leading to discrepancies between predicted and actual values. These discrepancies can lead to problems such as production planning failures and safety accidents. To cope with these problems, we developed a digital twin for real-time data acquisition and constructed a cycle time prediction model using a DNN. The proposed method provides two major benefits: (1) It can predict accurate cycle times by reflecting the robot's real-time condition; and (2) cycle time can be predicted from the robot program without stopping the equipment. Experimental results demonstrate that our proposed method outperforms both commercial offline programming (OLP) systems and experience-based methods-time measurement (MTM) approaches.