Network Delay Forecast and Master–Slave Consistency Enhancement for Remote Surgical Robots

摘要

BACKGROUND: The inevitable network delay can directly impact the process of remote surgeries and affect the master-slave motion consistency, and sudden changes in delay can compromise surgical safety. METHODS: Firstly, real-time calibration of unidirectional network delays is performed. Subsequently, the network delay is forecasted with a real-time training parallel recurrent neural network for safety warnings, and the real-time forecast of slave manipulator position is performed to enhanced the master-slave motion consistency. Finally, the forecast accuracy across multiple scales is assessed to provide feedback. RESULTS: The programme can operate on standard computers at distances of at least 630 km. Our forecast method meets the real-time requirement, demonstrates strong generalisation capabilities and reduces the impact of network delay on master-slave motion consistency to approximately 20%-80% of its original level. CONCLUSIONS: The proposed forecast method enables real-time delay forecast for remote surgeries, reducing the impact of delay on master-slave motion consistency.