A real AI challenge would be to create a general purpose handyman robot that can do common household installation/repair/maintenance on things not explicitly designed with automated maintenance in mind. Cc
Conversation
When you poke at “automated” maintenance you realize that it’s mostly about designing for higher resolution asynchronous replacement rather than repair, often with diagnostics on subassemblies/components. This is cheating. Makes economic sense but finesses the AI challenge.
Replying to
I think the DARPA grand challenge on search and rescue robotics (how did that go btw? anyone track that?) is a step in the right direction. Addresses the analysis half of repair by navigating brokenness and messiness.
2
3
Also interesting to reflect on Asimov’s justification for humanoid robotics: that androids would make great space explorers because they’d have to modify alien environments in human friendly ways. Some version of that might be true.
1
3
Perhaps a generic repair-capable intelligence would have AI/robotics all in the spirit of “opposable thumbs”. Speaking of which there was a management book called “opposable mind” which I haven’t read that might have interesting ideas here 🤔 amazon.com/Opposable-Mind
2
10
