STITCH
Augmented Dexterity for Suturing Technique Involving Thread Coordination Handling
Kush Hari*, Hansoul Kim*, Will Panitch*, Kishore Srinivas, Vincent Schorp,
Karthik Dharmarajan, Shreya Ganti, Tara Sadjadpour, Ken Goldberg
UC Berkeley
* Denotes Equal Contribution
TL;DR: Augmented dexterity suturing pipeline for the da Vinci surgical robot. Paper
Overview
We present a Suturing Technique Involving Thread Coordination Handling (STITCH): a continuous suturing policy for the dVRK which performs needle insertion, thread sweeping, needle extraction with suture cinching, needle handover, needle pose correction, and failure recovery. To ensure each stage of the STITCH finite state machine runs smoothly, we develop a novel visual 6D needle pose estimation module using a stereo camera pair and an augmented dexterity suturing motion controller. We test the STITCH pipeline on a wound phantom and conduct experiments to compare success rates between separate pipelines utilizing proprioception, no-servoing, and STITCH. In physical experiments, we find that on average we can throw 2.93 sutures without human intervention and 4.47 sutures with human intervention indicating that STITCH shows promise creating an augmented dexterity suturing pipeline.
Video of 6 Consecutive Sutures with STITCH Pipeline
Note: The video has motionless frame sequences removed and runs at 20x speed.
The video showcases a sped up consecutive execution of each step in our pipeline: needle insertion, thread sweeping, needle extraction with suture cinching, needle handover, needle pose correction, and failure recovery when a gripper misses its target grasp point as shown in the second suture.
Questions?
Contact kush_hari@berkeley.edu to get more
information on the project
This paper is published at ISMR 2024.