ML-driven wireless CNC machine tool control: A concept to address latency and jitter in industrial networks

The increasing flexibility of manufacturing processes to fulfill different customer needs requires reconfigurable and interconnected manufacturing systems. Wireless sensors, actuators, and control systems are enablers in this evolution. They provide scalable and seamless integration of wireless machine controls into automated manufacturing systems, such as automated guided vehicles, robotics, and CNC machines. Wireless machine control increases productivity and resilience through networked machine-to-machine communications, by effectively utilizing unused computing capacity. However, implementing such networked control systems in industrial automation requires reliable and deterministic communication network performance, particularly in terms of end-to-end latency, jitter, packet loss, and availability. Despite the benefits of wirelessly connected cyber-physical production systems, the lack of time-deterministic wireless networks poses a significant challenge to the seamless migration of machine tool control due to its high sensitivity to network disruptions. This paper proposes a concept to overcome this challenge by implementing a machine learning-based latency prediction model into the CNC machine tool control to ensure that communication latencies do not affect control performance, thereby overcoming the limitations of current wireless network technologies.