This is the tank. One day, perhaps it'll earn itself a real name.
Being my first robot, it is of course a skid-steering obstacle avoider. Since I already had an Arduino and wanted to continue with it, I didn't follow the Start Here robot, but instead pretty much copied LarryBot. It is made of: a Tamiya 70097 twin-motor gearbox, a Tamiya 70100 track and wheel set, an Arduino Uno and an SN754410 H-bridge chip.
Unlike LarryBot, I am currently using the motors that come with the gearbox (and thanks to those on the forums, especially OddBot, that helped me think my way through powering their comparatively puny voltage (see here)). The recommended RM3 motors are on back order, but I got impatient. I have found that, at least with these motors, 4 NiMH AAs are easily enough - these put out 5.3V between them. Not pictured is the 9V battery that is carried behind it - the chassis as it stands was wonderfully quick to throw together, but far from satisfactory. I think it needs to grow somewhat. I may take a cue from real tanks and have it rather taller than its tracks.
Notable in this one-of-a-kind construction is that nothing is actually fixed together. The foamboard is cut to press together, the wheels hold it together. It's not exactly solid, but I am at the early prototype stage, and I'd rather face the hassle of re-attaching wires than re-soldering them. As it stands I am not happy with this shape - the only place for the breadboard and Arduino is on top, and there is not enough room for a planned turret, let alone having the turret's base servo sunk into the top panel (as here). I don't have much access to fancy powertools and so on, so I am reticent to make a more permanent chassis until I have the prototype down. Sadly, this will be difficult when I get to the turret - not a clue how to put that together.
Of course, there is no turret. There is no ultrasound sensor. Because I have hit a bit of a snag. One set of tracks is pretty much useless. It runs perfectly when unladen, but against any resistance it stops. That's 'one side', not a particular side, because if I swap the output jumpers, the other side fails. After a while I put two and two together. Earlier on I ran it without grounding one side of the H-bridge. It got really rather hot before I noticed. That's the side that's now failing. It seems my first attempt at my first robot has ended with me frying a chip.
So, a lesson for anyone as unobservant as me: Always ground your H-bridge.
Still, it's been a learning experience. I'm still waiting on the better motors, so I suppose the time to order a new H-bridge won't kill me. In the meantime, I suppose I will try to think of methods of construction. And perhaps work out a way to make my own plugs and sockets for attaching, say, the motors and the ultrasound sensor more securely but still able to be hot-swapped around.
I have no idea how other people figured out how to space their wheels for this trackset. The closest thing I found to documentation was the patterns on this Instructable. I adapted the spacings to make it symmetrical, and have attached a zipped Inkscape svg in case that is of any use to anyone in the future.