A shared control method for safe teleoperation of an immersive-telepresence robot

Abstract

This thesis presents a shared control framework for telepresence robots to balance the user input with autonomous safety, particularly in obstacle-rich environments. While many telepresence robots support autonomous navigation, operators often need to maintain control over the robot’s path to adjust for situational preferences not encoded in navigation algorithms. This thesis proposes a Dynamic Window Approach (DWA)-based control system that integrates user commands directly into the robot’s trajectory scoring mechanism. Unlike previous approaches that rely on global paths or sacrifice user intent for safety, this system enables real-time obstacle avoidance without overriding user intent while automatically realigning the robot to the user’s original intended path after obstacles are avoided. The proposed solution was deployed on a ROSBOTXL platform equipped with a 360° camera, providing immersive visual feedback through a virtual reality headset. This setup enables natural viewing of the remote environment while the shared control system mediates between following user inputs and autonomous obstacle avoidance. To evaluate the effectiveness of our approach, a user study was conducted with (n=12) comparing the proposed shared control method against conventional manual control. Results indicated that users traveled farther with shared control (M=22.20m, SD=6.20m) compared to manual control, in which they also needed to avoid the obstacles, (M=18.24m, SD=5.13m). However, this increase in distance was not statistically significant. Workload measures showed a trend toward lower cognitive burden with shared control (M=50.36, SD=17.23 vs. M=56.31, SD=15.19), and 66.7 % of participants found the shared control method easier to use. However, preference was more evenly divided, with 58.3 % preferring shared control. Qualitative analysis revealed that users who preferred manual control valued the feeling of direct authority, while those favoring shared control appreciated the reduced cognitive load and autonomous obstacle avoidance capabilities. The findings suggest that DWA-based shared control can enhance telepresence robot operation while preserving sufficient user agency, particularly in scenarios requiring navigation around obstacles. The study also provides insights into the trade-offs between manual control and shared control in immersive telepresence applications.