Work

Multi-National Validation of 5G-Enhanced Teleoperation with AR Interface and LSTM Prediction for Flexible Robotic Endoscopy with Sub-500ms Latency

Year: 2025

Type: conference-paper

Abstract: Remote robotic-assisted teleoperation is vital for overcoming geographical barriers to flexible endoscopy. To address the requirements of both flexibility and high precision in telesurgery, an intuitive robotic-assisted system for remote control is implemented. The system utilizes 5G Customer Premises Equipment (CPE) devices, combined with the WireGuard protocol, to establish a secure, low-latency communication link. Through this connection, the remote surgeon controls the endoscope via a Unity-based interface enhanced with Augmented Reality (AR). Furthermore, a trained Long Short-Term Memory (LSTM) model predicts the tip’s position based on motor pulse inputs, improving control accuracy. The long-distance experiments conducted between Shenzhen, Shandong, Switzerland, Dalian, Ireland, and Hong Kong showed that the image transmission delay ranged from 50 to 200 ms, the control delay was approximately 100 ms, and the overall delay was about 300 to 400 ms. In addition, the LSTM model achieved a prediction error predominantly within 5 mm. These results validate a low-cost, high-precision remote endoscope control system with acceptable latency, underscoring its potential for practical telesurgery applications.

Source: Procedia Computer Science