Coming up next: Hammer and Head

So, my next robot project has two robots: Hammer and Head. Hammer will have a hammer of some kind and Head will have a big head. Hammer and Head are going to have some fun together. Well at least Hammer will have fun...

Here's a list of updates to help with page navigation (if it ever works):

• Update 21 Jun 2010
• Update 22 Aug 2010
• Update 23 Aug 2010
• Update 24 Aug 2010

And now on to business:

Basically the idea is that Head will just wander around and Hammer will try to catch Head. I'm planning to "cheat" a little bit so Head will have an IR beacon to help Hammer locate it. My plans don't anything more except at least Hammer will have four wheel drive. That's because bad guys always have 4WD don't they. Although really bad guys have 6WD ;-) Oh, and Head will probably be running on Picaxe and Hammer on some AVR chip me thinks.

Because I'm doing this for learning and fun that's enough planning for now. Good way to learn is by mistakes and mistakes usually happen when you don't have good plan so I just started building Hammer. I already had some Solarbotics GM9 gear motors, wheels, sprockets, chain, paint stirrers, hot glue and other stuff. That equals a nice chassis for Hammer.

Here's what I managed to put together yesterday and today: Chassis for Hammer. There's some pictures of it below and short descriptions of what I have done.

Hammer's rear wheel attachment

All Hammer's wheels are red GMPW wheels. Rear wheels are attached to GM9 motors with a M4 screw that goes through the wheel, 3 washers and a M4 nut (for spacing, can't see those in the picture above), a sprocket, 2 washers and two pieces aluminum tube. I drilled GMPW wheel's center hole with 3,5mm drill so M4 screw fits tightly. In addition I used a drop of super glue to secure the wheel to the screw.

I used two pieces aluminum tube to attach the screw to GM9. The first piece is 8/6 tube (8mm outer and 6mm inner diameter). It's about twice the length of GM9's shaft. I flattened the other end of it hoping it would fit GM9's shaft just like that. Well it didn't but then GM9 met Dremel and my little piece of aluminum tube now fits on the shaft just fine. I drilled 1mm hole though the tube and GM9's shaft so I can put a pin (just a piece of iron wire) through to secure them.

The second piece of aluminum tube is cut from 6/4 tube and it's about half the length of the first one. So the second piece fits inside the first one and M4 screw fits inside the second piece. I drilled a 1mm hole through both pieces and the screw, put a pin through the hole and that's it. Then I did the whole thing again and I had my rear wheels attached to motors.

Hammer's rear wheels and motors

Rear wheels ready, time to figure out how to attach front wheels. I thought they could be done pretty much the same way as rear wheels except there's no motors. I had to find a way to make them rotate freely. After some quick brainstorming (more like a one giant lightning flash) I did this:

Hammer's front wheel attachment 1/3

There's M4 screw, washers, a nut and a sprocket just like in rear wheels. The difference is that there's 2 pieces of 6/4 tube and one long piece of 8/6 tube. Both 6/4 pieces go on the screw and the one closer to the end of the screw is secured with super glue. The tube piece next to the sprocket is now held in place by the other piece of tube and it's rotating freely. The long 8/6 tube goes on the both smaller pieces. I drilled a hole in the 8/6 tube and put a screw through it so it's attached to freely rotating piece inside of it. Now I can attach the 8/6 tube anywhere I want and the wheel will rotate. Here's a couple of pictures of the whole front wheel thing to clarify things:

Hammer's front wheel attachment 2/3

Hammer's front wheel attachment 3/3

I hope those pictures give you the idea how I did it.

Now that I had wheel attachments ready I needed chassis. Paint stirrers and hot glue For Teh Win! I also used that metal ribbon thingy with holes in it. I really don't know what it's called neither in English or in Finnish. You can see it pictures below (tell me if you know what it's called :-). Chassis is pretty straight forward thing so I just post pictures right here:

Hammer chassis top

Hammer chassis bottom

Hammer chassis side

That's how Hammer looks like right now. For a quick test run I used Picaxe 28X1 + 28-pin project board to see how this thing moves. It seems to move quite nicely but turning is a little bit hard (I kinda expected that) but it still turns. I might make the chassis a bit shorter (rear and front axles closer together). I think that would help with turning.

I also thought that I could probable use some wheel encoders. Never done that before. Those GMPW wheels seem to have some kind of "almost" encoder markings. Maybe I'll just use a magic marker to enhance those. It's going to get a bit tight though because those sprockets and chains are a bit in the way.

Update Jun 21 2010:

Today I made the chassis of Hammer a bit shorter. I thought it might help with turning. I'm not really sure if it helped or not. It seems like it's turning a bit easier now. Well at least on carpet. Those GMPW wheels have a good amount of grip to turning is still a bit difficult on hard surfaces. I shot a video where you can see how it handles. Here's a picture of how the chassis looks like now:

Hammer's chassis after some sawing

I'll probably have to replace those GMPW tires or try to coat them with something that has a bit less friction.

Update Aug 22 2010:

Finally I got myself building again and managed to make an evil-eye-thingy for Hammer. Basically it's more or less like Oddbot's compound eye but without IR-leds. I'll place IR-leds later to Head so Hammer can track it. I used SFH 313 FA phototransistors and 47K resistors. I experimented with bigger resistor values too and sensitivity seemed to get better but I wasn't sure if it could cause noise problem so I decided to try with 47Ks first. Not really much more to say about it so here's some pictures.

Construction phase. I used some hot glue (what else ;-) to keep those phototransistor in place during soldering.

Some soldering done.

Finished.

Bottom side.

Next thing to do would replacing that evil-eye-thingy on top of servos and try to track an IR-led.

Update Aug 23 2010:

Oh yeah! I'm on fire! Two updates within just few hours! I just couldn't leave that evil eye alone and go to sleep. I had to try it. So I quickly assembled some "eye muscles" from two Tower Pro SG90 micro servos and attached them and the eye to Picaxe 28x1 (eye will be run by ATMega8 later). I already had Picaxe 08M doing PWM with an IR-led from my previous tests with IR-phototransistors so it could play the role of "the bait". (PWM was really a simple modulation. My Picaxe test code doesn't take any advantage of it.)

IR-led tracking seemed to work, at least somehow. There seems to be some differences in phototransistors' sensitivities and that was expected. However the difference between left and right was quite notable. It might be that I have accidentally damaged one phototransistor on left (they work in groups of two).

Below is a picture of the (ugly) setup and there's also a video showing the system in action (also on Vimeo).

Evil Eye test setup

And here's the simple test code:
b0 = 140 ' Smaller = right
b1 = 150 ' Smaller = down

servo 0,b0
servo 1,b1

main:
    readadc10 0,w1
    readadc10 1,w2
    readadc10 2,w3
    readadc10 3,w4
    'sertxd ("Left: ", #w1, ", Right: ", #w2, ", Up: ", #w3, ", Down: ", #w4, CR, LF)
    if w1 < w2 and b0 > 75 then
        b0 = b0 - 1
    endif
    if w1 > w2 and b0 < 225 then
        b0 = b0 + 1
    endif
    if w3 < w4 and b1 > 120 then
        b1 = b1 - 1
    endif
    if w3 > w4 and b1 < 225 then
        b1 = b1 + 1
    endif
    servopos 0, b0
    servopos 1, b1
    goto main

Update Aug 24 2010:

Just got the evil eye thing running on ATMega8. There's video available on Youtube and Vimeo. Code is AVR C code written/compiled on AVR Studio/Win-AVR. Basically it just reads 5 ADCs and turns servos according to ADC readings. Servo control is done using one of ATMega8's 8-bit timers. ACD readings are also dumped out from UART for debugging purposes.

The eye is a bit twitchy at the moment. I did a little bit a debugging and I'm getting nice and stable readings when I hold an IR-led if front of the eye. When I move the led away from the eye readings decrease steadily for a while until they drop down drastically and the eye starts twitching. Curiously it looks like (tried only 3 times so far) that when readings drop they always have same kind of pattern (lowest and highest values always come from same phototransistors).