Confidence-Based Intent Prediction for Teleoperation in Bimanual Robotic Suturing
Zhaoyang Jacopo Hu, HAOZHENG XU, Sion Kim, Yanan Li, Ferdinando Rodriguez y Baena, etienne burdet
AI summary
Problem
Traditional robotic teleoperation suffers from control delay, sensory limitations, and stability issues, while fully autonomous systems lack the adaptability needed for unstructured surgical environments.
Approach
The authors developed a shared control framework that combines a Transformer-based gesture recognition model for high-level intent prediction with a confidence-driven controller that dynamically adjusts robot assistance based on real-time sensor reliability.
Key results
- Real-time Transformer-based surgeme classification from kinematic data
- Confidence-driven Intention Assimilation Controller blending human and robot control
- Custom CHENA needle alignment device for improved manipulator handling
- Statistically significant reductions in task completion time and improved user satisfaction
Why it matters
Provides a practical pathway to safer, more efficient robotic surgery by bridging the gap between manual teleoperation and full autonomy through adaptive intent prediction.
Abstract
Robotic-assisted procedures offer enhanced preci- sion, but while fully autonomous systems are limited in task knowledge, difficulties in modeling unstructured environments, and generalization abilities, fully manual teleoperated systems also face challenges such as delay, stability, and reduced sen- sory information. To address these limitations, we propose an interactive control strategy that assists the human operator by predicting their motion plan at both high and low levels. At the high level, a surgeme recognition system is employed through a Transformer-based real-time gesture classification model to dynamically adapt to the operator’s actions. At the low level, a Confidence-based Intention Assimilation Controller adjusts robot actions based on inferred user intent and shared control paradigms. The system is built around a robotic suturing task, supported by sensors that capture robot kinematics and task dy- namics. Experimental results across users with varying skill levels demonstrate the effectiveness of the proposed approach, yielding statistically significant improvements in task completion time and user satisfaction compared with traditional teleoperation.