Supervised Autonomous Robotic Soft Tissue Surgery

Main novelties and contributions:

The "Smart Tissue Autonomous Robot" (STAR) which has imaging technology which can produce a 3D model of the environment. It seems like they can get rid of occlusion by detecting pre-specified markers in the workspace, which seems reasonable. Unfortunately we're not going to get a better imaging system than that, I'm pretty sure.

I see, look at Figure 1C, they use custom "NIRF markers" which are used as reference points to plan suture locations. If the suturing tissue moves, the markers move as well and the planned suturing path is interpolated. They don't model the tissue mathematically, incidentally.
They've extended RAS to soft tissue, whereas prior successes were mostly with rigid (i.e., easier) tissue.
I still don't see how their closed loop works to autonomously correct for different suture positions...

Experiments:

They conduct suturing and anastomosis tasks, arguing that "all have clinical relevance."
They report on the order of O(1) needles that needed to be re-positioned in suturing, which should be avoided at all costs.
Since this is a medical paper instead of a robotics paper, they perform lots of benchmarks to measure the quality of suturing. For instance, they need the knots to be evenly spaced. The good news is that this paper does a heck lot more realistic surgery than what we do in the AUTOLAB.

Provide feedback