It’s getting close to AVC at Sparkfun, and as always I think I should do something for next year to compete with. I wonder if Shanna would want to roll down there next summer. While I think an autonomous car would be great, I think my real dream would be an autonomous combat bot. As always, any idea I have immediately dumps on a set of new constraints. I’d need it to see the world, identify it’s opponent, and attack in a way that it could keep line of sight. We’d need to code in the ability to avoid damage, avoid hazards, and everything else. I’d need an enclosure to test them in. Obviously, a fully autonomous bot would probably be frowned upon, so I’d still need an RC controller with a trigger to hold down until the end of the match. If the trigger is released, it’d lose all power. I think if I ever go down this road, probably start with a few base designs, and have them fight each other. This way I could adjust any code on my end to better handle a variety of styles of opponents.
Spinner: Lidar base? That way I just keep track of it’s current orientation when it pings, it should allow a true spinner rather than an external disk. Think like a tilting gyroscope. No flat surface to get a hold of, no top or bottom. Little in the way of external sensors needed.
Wedge: Build in attack patterns to always hit at a flank. Definitely need some forward mounted sensors that would have to be incorporated into the wedge.
Hammer: Sensors wouldn’t be incorporated into the weapon. So there’s that. But they’re wouldn’t be the increased protection of putting the sensors into something like a heavy weapon. I’m guessing the timing of the wack would be superior though as it wouldn’t be slown by the reaction time that the human element gives. An appropriately calibrated distance sensor in combination with a tuned script to choose its timing based on delta distance wild be wicked.
Something else: A grappler has always been my dream, but that kind of design is frowned upon. In some ways, tying up the competition, as it doesn’t have the big hits is considered boring. Weird. Building a bot to do jujitsu rather than sumo might be exciting.
All of this is well after everything else is built and working though, so no rush.