Core Concepts
STITCH, an augmented dexterity pipeline, performs Suture Throws Including Thread Coordination and Handoffs using novel perception and control techniques.
Abstract
The STITCH pipeline automates the surgical suturing task by iteratively performing needle insertion, thread sweeping, needle extraction, suture cinching, needle handover, and needle pose correction with failure recovery policies.
The key components of the STITCH pipeline are:
6D Needle Pose Estimation Module: STITCH uses a combination of deep learning, analytical, and sampling-based approaches to accurately estimate the 6D pose of the surgical needle. This enables closed-loop control of the needle motions.
Augmented Dexterity Suturing Motion Controller: STITCH coordinates the various sub-tasks of the suturing process, including needle insertion, thread sweeping, needle extraction, suture cinching, needle handover, and needle pose correction. Novel motion primitives are introduced to improve reliability.
Motion Failure Recovery: STITCH includes recovery mechanisms to handle failures during needle extraction and handover, retrying the motions up to 5 times before declaring failure.
In physical experiments across 15 trials, STITCH achieved an average of 2.93 successful sutures without human intervention and 4.47 successful sutures with human intervention. The pipeline demonstrates the potential of augmented dexterity to enhance surgical capabilities.
Stats
STITCH achieves an average single-suture success rate of 69.39% and a mean sutures-to-failure of 2.93 over 15 trials.
With human intervention, STITCH achieves a single-suture success rate of 83.33% and a mean sutures-to-failure of 4.47.
Quotes
"STITCH iteratively performs needle insertion, thread sweeping, needle extraction, suture cinching, needle handover, and needle pose correction with failure recovery policies."
"We introduce a novel visual 6D needle pose estimation framework using a stereo camera pair and new suturing motion primitives."