Suturing Tasks Automation Based on Skills Learned from Demonstrations: A Simulation Study

Overview

Implementation of imitation learning algorithms for automating surgical suturing tasks in simulation.

Key Contributions

• Implement imitation learning algorithms for suturing automation in simulation, achieving 95% success rate for task completion, task generality on the order of 91.5% and 20% less task execution time
• Develop a novel pipeline for synchronized data collection and conduct user study for human demonstration acquisition using the physical dVRK
• Data subsequently utilized for building a 2024 NeurIPS-published dataset to support surgical policy learning