I second robots; but more like "waldoes", generic amplification of
my human capacities, on a bigger-than-me (backhoe), stronger-than-me
(force amplifier to assist my neuromuscular degeneration), or tinyier-than-me (microfabrication) bases. Make them available and plentiful.
A language translation system like Reasoning, Inc's "Refine", but intended to swallow code from ANY language,
including it's
libraries, factor and globally optimize, and emit code in
any other language.
With some tool that points out "routine A" is a not quite complete (errors or
omissions) implementation of "routine B".
Eventually, using the accumulated library of code, produce a goal (constraint) based
language that asks interactive questions, to produce new programs.
These are some great ideas, thanks! I've seen a few intuitive programmable robots (things like a car that does different things depending on the color it drives over), but these are some I hadn't found yet.
Maybe think about building separate robots for each task (tub, etc...) that can be permanently attached to wall or the ceiling near its work spot. Tasks could be accomplished using special arms for each sub-task. A lifting arm, a chemical spraying arm, etc...
It would take the "where am I" "where is the target" out of the robots equation. You could simply hardcode the desired movements into the robot. Then if you wanted to sell them you could build a learning routine into the code so the public could easily use it after bolting it to the walls in their houses.
In there is a note that Wolfram is working on, or at least thinking about, the big unsolved problem - useful robots. Robot manipulation in unstructured situations is still very poor. But maybe someone will figure out a way to apply newer machine learning techniques to that. Google had a research group working on that, but they haven't been heard from in years.
The approach of the large language model, where you have a huge training set of general purpose info and a small prompt for the current task, might possibly work.
I've never heard of Robot OS, looks interesting! Yea, unfortunately its the niche stuff like that which experience the most problems (or at least I assume its fairly niche - I do nothing with robotics).
In those cases, one can, without too much effort, write forward-compatible python 2 code at least.
- The icon on the forehead of the robot should indicate who is currently controlling it.
- This could even be a little screen for a webcam in front of the controller.
- An office full of these robots could be an innocuous setting for a "secret" Turing test.
Maybe there's room for improvement, a software guided robot could be faster, more efficient, cheaper to run and occupy less space than clunky old machinery.
This is exactly the problem DARPA NOMARS seeks to solve.[1] If there are any strong roboticists here on HN interested in working in this domain, PM me.
reply