Thanks Jeff,
I hadn’t realized that. It does work out about $5 per battery cheaper to raid a DeWalt pack.
Thanks again!
Its been a long time between updates. This has largely been due to my frustration in finding an electrical fault. 120 hours of testing, retesting, rewiring and not being able to figure out why all of my controller stamps were malfunctioning (2x BASIC, 1xCorridium, 1xAtom). Eventually, I noticed one of the little copper sleeves inside the stamp socket on the bot board was missing, and one pin wasn’t making contact.
I didn’t touch LEX for a long while after that. However, he’s been staring at me since, and so… There is now an update.
LEX is uglier still.
Actually, he was creeping me out a bit looking like that, especially when colour tracking, so I added a faceplate (very thin shim of black plastic). My wife wants to add eyebrows at this point.
The servos are over-volted to 6v6 and he has no trouble walking around - I’ve not seen any magic blue smoke yet, and have been keeping an eye on temperature readings. Hardware changes this iteration are fairly extensive. SSC 32 was upgraded to V2, all the wiring was redone. A 15 amp mini-blade fuse was added, visible just in front of the ‘backpack’ (SSC, BotBoard, OLED display).
The head now has a CMUCam 2+, EzSonar and IR sensor. The pan-tilt servos are run directly from the CMU2+ for tracking purposes. The CMU is also serially controlled by the Coridium Lite stamp. Because there is so much in the head area (speech synth & speaker, CMUCam, sonar, IR), I ran 6 strand ribbon cable from the BotBoard to a custom daughter board to make the cabling a little more sane.
The daughter board handles power distribution, I2C bus, voltage dividers (Coridium stamp, while 5v, has 3v3 A/D) and breaks out serial comms for the camera (far easier to have the ribbon terminate in one spot and run seperate wires). Servo power for the CMUCam is provided directly from the batteries, which can source far more current that I’ll ever need on this guy.
Surprisingly enough, he’s quite well balanced front to back and left to right, but moving the head whilst in the middle of a step can cause some instability. Testing has all been done with PC control, as I need to rework the software extensively.
sorry if i missed this in an earlier post but, how are you controlling the scout… PS2?
good to hear you sorted out the microcontroller socket. what chip did you settle with… Atom Pro?
im working on a scout like project and im having prolems calculating the IK/FK for real-time control. are you anywhere near this?
nice project. 8)
He’s autonomous, no RC (for dev work, he’s slaved up to serial on a PC, which is about where I’m back to now). He was walking around and not bumping in to things with the IR/Sonar. I went with the Coridium ARMexpressLite as a controller. It has (or had, I haven’t checked in a while) a much nicer C library than the Atom Pro.
No inverse kinematics at all so far. I added a memory chip to the SSC-32 and all the walk sequences (squat-to-stand, walk forward, turn left, turn right, etc) are stored in there, to be called at need. I’d love to get into IK for him, but there’s so much to do before I get to that, I’m happy with static gaits for now.