I've developed an automated sorting assistant using industrial robotics,
computer vision, and machine learning.
It's been some time since my last update, and what a whirlwind it's been!
Last October, I believed I'd hit the target of 90% object identification
accuracy across 58 part numbers and 6 object types. However, stress testing revealed
persistent problems, with some object types falling short of the 90% mark, halting
deployment plans. Check out the details on my blog ( https://www.thirdarmrobotics.com/tarblog.html ).
I'm thrilled to share that, after significant upgrades over recent months,
the system now consistently achieves over 90% accuracy for all object types and
part numbers it can identify. With this milestone reached, I'm ready (once
again!) to push forward with deployment.
Please email me (redacted) if you need more info and/or want
to be contacted when deployment begins.
I’ll do my best to answer any questions or respond to any comments you might
have.
USA, Arizona
Canada, Ontario