Its about time I post some images of ‘LEX’, the scout I’m working on.
All programming has been done in C. Trig functions are calculated at runtime (using transcendentals) rather than using a lookup table or the GCC math library.
There’s a few home brew circuit boards in there for some basic functions. I2C power is provided by a 9v battery, switched with a transistor from the servo batteries. There are also some voltage dividers as the ADC needs a 0-3v3 range.
Walking isn’t all that good at the moment (yes, he is a bit nose heavy), I need to spend a lot more time working with the Visual Sequencer on that.
I’ve been having a lot of fun with the display and scanning, and have a few other things I’ll try with that before moving onto the walk cycle.
I’d like to combine the two display outputs and filter them a little more, and possibly look at correcting distortions produced by the pan/tilt sweeps.
Thanks for the fantastic kit, Jim, I’m having an absolute blast working on it.
(I hope you don’t wince too much when you see all the things bolted on to it…)
NiMh, actually. 3 on one side, 2 on the other (and a small counterweight). I liked the look of the 3xAA holders, but didn’t care for the internals… They’re completely gutted inside and the batteries are soldered together, with 18AWG running to the SSC32. LiPos would have been better, but I have never used them before and am a bit wary of them.
I’d love to be able to mount a CMUCam 3 on the pan-tilt head. The wiring is all in place for it…
HS-5645 servos in all leg joints. 422s for the pan/tilt.
I can’t give you an exact weight of everything (scales I don’t have). I can say that most of the weight in front comes from the two servos running pan/tilt.
The sonar and IR are mounted on a small square of lexan.
There is a small circuit board on each side of the pan servo, one being an I2C bus, the other being power distribution and some simple voltage dividers. On top, mounted behind the main power switch (lashed on to the mount for the SSC-32), there is a 9v battery. The rest is wiring. The mesh covering the wires is light (and overused, but I love the stuff).
The Devantech speech synth unit is slung under the main chassis.
The display itself, whilst making it look heavy at the back, doesn’t weigh much. The display housing is lexan sandwiched between two sheets of polystyrene sheet. Most of the lexan is cut away behind the poly, and it weighs less than the lexan used to mount the SSC-32.
Very nice! Awsome! I especially like the braided wire looms you used. I see a color display on the backpack, what is generating the color display graphics? What kind of display is it? I know there is an EZ color Display you can buy that will generate color graphics from a basic stamp II. Are you using a BAP or a Basic stamp?
You need to at least write up a tutorial on how you got the display to work with your setup. This is the first time I have seen it done on these forums.
I’m using a Coridium stamp to run this, though I’ve run it from an AtomPro and Basic stamp in the past. It doesn’t really matter. The smarts for the display are in the display.
It is an old model from 4D Solutions. It’s an Australian company that sells through Dontronics.com (I purchased it while I was back there last year, not sure of US distribution). It has 160x128 resolution, full colour, and has 8mb of memory for storing various graphics (which I haven’t investigated - I lost my RS-232 to TTL converter in a cleaning incident).
It is (at least the one I have) a devil to mount.
But, the best thing about it is that it is serially controlled… There are built in commands for text, pixels and line plotting (including circles).
That is exactly how the EZ displays work, they have built in functions so all the Stamp has to do is send serial commands to make the display plot pixels, lines, set colors and shapes, text etc…
They do whine a little, but not more than I expected. This is my first robot, so I don’t really have any experience with servos under load to compare them with. When he’s standing they’re fairly quiet, when he’s lurching about… well, they’re louder, but not so loud that it annoys my wife/dog/cat/neighbours.
They don’t seem to get really hot either, but I haven’t run him around for an extended time. Also, while he’s not actively exploring, he’ll go into a crouch where all the weight is supported by balancing his hips on his feet. The servos are silent at that point.
I also do have four of the load balancing springs attached to help out where the most strain seems to be, they definitely made a difference.
I think the images are a bit deceptive, though, I don’t think he’s overloaded (yet - I’m definitely working on that…). There isn’t much weight at all in the display backpack (its lighter than the SSC-32 according to my rough-n-ready balance measure).
The two servos in the nose for pan-tilt are the definitely the heaviest add on, and they weigh more or less the same as a 5xAA battery pack.
At this point, I’m working on a better gait, then it will be software tinkering. I still want to add an CMUCam, I’d be happy to sacrifice the tilt servo for that if becomes too much weight.
The 5645s are some tough servos. Any servo moving is going to make noise especialy onder a load. There are, after all, motors and gears running.
Ill be watching you project, it very interesting and well done. For a first bot–you would never know it was your first bot by looking at what you have done so far.
I apologise for the (huge) delay in updating - I got sideswiped by life and I’ve only just got back to my robot and the forums.
There have been a few changes to LEX since I last posted (and if you thought he was a bit ugly before… he got worse).
I had a few problems with the speech synth unit, it has been removed, freeing up quite a bit of space. I’m hoping to get it back on at some point, but its not a priority.
The rechargeable AA batteries I was using just weren’t cutting it. Fully charged, he’d take 4-5 steps and then spasm and fall over. After that he’d flail like ED-209 in a stairwell. He’s now running on a pair of A123 batteries @ 6v6. He’s waddling about quite nicely.
I’m being a little conservative with his run times at the moment, with the servos running a little over the rated 6v. I’ve also added a fuse (with 60C burst discharge, I wanted something that would melt before anything expensive if anything should happen to short).
That is one sweet looking scout Matt! I have (had) one myself, it’s in pieces right now (cannibalized it for the servos), but I remember when I had it, I was considering to stick some load balancing springs too, I couldn’t tell where you implemented springs in your pics. Did you place it at the two (reversed) knee joints? I found those two servos were carrying the most load, especially the single one, when it is momentarily lifting one leg and supporting all the weight on the other to walk.
I never got it to successfully walk since I discovered that my servos (645MG, the analog version) was way underpowered for it. I think for the scout, since it has reversed knees and with a slight crouch-stance like a chicken, it requires a much stronger servo, more likely the 5990TGs with its 333oz-in of torque at 6.0V…
Funny that you mentioned the AA-pack. I was using a 5-cell AA pack that I built myself out of the Energizer’s NiMH 2500mAH cells (which were horrible to begin with) and experienced the same problem. It would lift one leg and then spasm and just fall over (luckily, I was no where near a staircase ).
Although I wouldn’t recommend it, before I went the AA-cell route, I had some matched 6-cell Sub-C packs (for my R/C car), which has an open-circuit of about 8 to 9V, and this thing were like steroids for the servos. They heated up within minutes of running, but power out of the “wimpy” 645MGs were plentiful.
I think when I eventually get back to the scout, I would like to entertain a 5-cell Sub-C solution and outfit the scout with 12 5990TGs…
How do you like the A123 batteries? I think for payload consideration, a lipo pack is in the order. With bipeds, it is a challenge, you have to have a battery that have great discharge characteristic and be light weight…
Thanks! In those photos (which are a bit old) the springs had been removed. I did have them placed at the knees, as you say, and also at the front of the ankles. Latest pictures are here (taken with a cell phone in poor lighting, I’ll get better ones and a walk video soon). I should have shrouded the noodle-wire in the same cable mesh, but that’s hindsight for you…
I’m not 100% happy with his walk cycle as yet (its more of a crouched shuffle), but I think that should be correctable with a few tweaks to the servo positions. Even with the shuffle, I was very happy to finally see him walking consistently.
The A123s are fantastic, I can’t say enough good things about them. So far the 5645s have been happy with the 6v6, and the batteries provide plenty of amps. I’m still fairly cautious on run times, not had him walking for more than 10 minutes at a time (and keep testing them with a thermal probe - I’m sure if robots get tired, he’s tired of the probe). The only mild draw back to them is that they aren’t flat packs.
I actually purchased the six-pack developers kit direct from A123 and a Hyperion EOS0606i charger from MaxAmps (thinking I’ll use more A123s on future project, the charger supports balancing up to 6 cells). I wired up two in series in a saddle pack arrangement, one cell on each leg (where the old NiMh packs were).
They have about a 15 minute charge time, far lower risk of fire than standard LiPo batteries, and 2.3Ah between charges. The two of them weigh in at about 5oz (plus wiring and connectors).
Hey Matt, I just wanted to say that your Scout looks awesome- I just started one of my own last week. In case you don’t know you can obtain 10x A123 batteries in a DeWalt 36V pack which might be a little cheaper than buying the developer’s kit. I’m currently using a LM 6-cell pack but I have 2x LM 5-cell packs in the mail and should be here tomorrow. If I need more juice later on I’ll probably upgrade to 10x A123 or GP3300 in 5 cell series, 2 packs parallel.
).