#robotics Logs

Dec 20 2018

01:13 AM rue_shop3: mechanical computer?
01:17 AM rue_shop3: might be large
01:42 AM rue_mohr: https://i.pinimg.com/originals/6d/93/fe/6d93feaaecfd603c37e407d2e58b7527.jpg
01:42 AM rue_mohr: I suppose thats a cute thing you can do with flat belts
09:14 AM veverak: hmm
09:14 AM veverak: I have octree representation of environment
09:14 AM veverak: where voxel has propability that there is something in the voxel area
09:15 AM veverak: so close to 0% means that there is air and close to 100% that there is something
09:15 AM veverak: I wonder how to effeciently mark 'close to 0%' which marks surface of the 3d model
09:16 AM veverak: I think that 'any voxel close to 0% that is above (on z axis) voxel which is close to 100% shall be marked 'surface''
05:22 PM ldlework: https://twitch.tv/roboduels if you wanna watch
09:01 PM syntaxfree: so about 10 years ago I played a little bit with Arduinos and servos with wires and breadboards.
09:01 PM syntaxfree: I had one of those things that's like a protoboard in that it has printed circuits following the same layout but it's full of holes for soldering?
09:02 PM syntaxfree: I hadit mounted to a servo with a bunch of sensors and some logic to try to balance the light or the heat on both sides of the board. fun stuff.
09:02 PM syntaxfree: so I'd like to make a robot on a hobbyist "chassis" that comes already mounted with wheels and motors; I've seen them for sale.
09:03 PM syntaxfree: but what I want to do now is to try deep q-learning on something like a beaglebone or raspberry pi. teach the thing to avoid magnetism or heat.
09:04 PM syntaxfree: does this make sense at all? I can run heavy training on a digital ocean droplet or something.
10:54 PM jadew` is now known as jadew
11:25 PM Malvolio is now known as Guest92374
11:59 PM SpeedEvil is now known as Guest50869