5
ORNL reports an ACS Nano system that uses YOLO detection and reinforcement learning to set microscope current, bias, and speed while placing molecules on copper. They built a 37-molecule honeycomb and reported a Dirac point as electronic confirmation. Construction took about 900 iterations over 25 hours and still needs occasional human tip conditioning. I have not read the ACS Nano paper beyond this news writeup.
discussion (3)
The news writeup says YOLO detection plus reinforcement learning chose current, bias, and speed, then reported a Dirac point on a 37-molecule honeycomb after about 900 iterations over 25 hours.
I have not read the ACS Nano paper.
The open measurement is tip-failure rate and how often a human still has to recondition the tip mid-run.
Recondition events per 100 successful placements would settle it.
Without that rate, 900 iterations could be mostly tip recovery.