dusting it off

it’s been a while.

my original plan with this blog was to write something every day. that clearly didn’t happen.

anyway, this weekend i revived an old project.

my SO-101 robot arm had been sitting around collecting dust for a while. the last time i used it, it could do teleoperation. i moved one arm, the other copied it. cool, but that was about it.

so i dusted it off, found the correct power supply after more searching than i’d like to admit, reinstalled environments, fixed the usual collection of things that stop working when you haven’t touched a project in months, and got it running again.

this time, though, i wanted the arm to do something on its own.

the current experiment is chess.

right now, it’s still basically a pick-and-place problem. look at the board, find a chess piece, move the arm toward it, pick it up, and eventually place it somewhere else.

“eventually” is doing some work there.

i can get the arm to move based on what it sees in the scene, which already feels like a pretty big step from teleoperation. but there are still a bunch of things i need to figure out:

• gripping a chess piece accurately
• actually moving the piece without knocking everything else over
• generalizing it so it can pick any piece from any square and place it on any other square
• figuring out when it should make a move
• figuring out whether this can mostly be solved with a flowchart-like “if this, then that” approach, or whether some part of it really needs a neural network

the eventual goal is to make the arm play chess.

maybe against me.

maybe against another robot arm.

i don’t really know yet.

and that’s kind of the point.

there isn’t a practical reason for me to build this. i’m not trying to turn it into a product. i don’t have a deadline. i don’t even know exactly what the final version looks like.

i just like making things, especially when i don’t completely know how to make them yet.

sometimes an old project doesn’t need a better purpose.

it just needs to look interesting again.