Archive for February, 2008

 

PID controllers!

Feb 28, 2008 in Bots, MIT, Bostoncaster, Cambridgeshire, Stuff

I just realized that some of the work we have been doing in 6.01 with state machines is getting pretty close to forming parts of a PID controller. I have been curious about those things, and have a grasp of how they work, but have yet to actually implement one to solidify said grasp. I suspect it might be coming up soon, since it is a very common control scheme.

This would be pretty neat for experimentation, as the only autonomous bots I’ve built to date have used proportional control only with an enormous deadband to avoid oscillation, or some sort of 3/8ths-ass fuzzy logic (not even half-ass).

Bets on when I’ll send one of the lab robots through the wall?

What the hell, people?

Feb 27, 2008 in Stuff

Equals Zero has been hit with a barrage of spam over the past week or so. It suddenly went from nothing to about 20 per day, fortunately all caught in the spamtrap.

Yeah, I know, still no Motorama event report… Consider it a vector sum of no motivation, busy classwork, and other projects. It’ll be up soon, I promise.

SPAAAAAAAAAAAAAAAAAAAAAAM SPAAAAAAAM SPAM SPAM SPAM...

Let’s see how many spam messages this one gets.

Motorama 2008: Saturday and Sunday

Feb 18, 2008 in Bots, Events, Project Build Reports, Test Bot 4.5 SP1

That was an insane weekend. Currently, though, I need to attend to a week’s worth of classwork that I need to make up, so won’t be writing a full event report yet.

Watch this space.

Motorama 2008: Friday

Feb 16, 2008 in Bots, Events, Project Build Reports, Test Bot 4.5 SP1

So, I wrote the last build report Thursday night in a hotel room in Harrisburg. Friday was 150g and antweights day, as well as early safety/checkin for anyone who wishes. There were plenty of larger bots being built in the pits.

Of all the things I forgot to bring, it was my camera’s (sucky proprietary) cable, so no pictures for now. I wasn’t able to secure an arm controller, but did get the following done…

  1. Make a arm restraint/guard to follow the rules
  2. Make an LED power light. It mights up the back end of the robot as in every version of TB since 4.0
  3. Make the charging plug, which worked fine.
  4. Install all the hardware. I only put on a bit of hardwareduring the build so each test didn’t involve putting in 30 screws. The bot’s fully loaded.
  5. Dropped a 3200uF capacitor on the 5v line. Spektrum receivers apparently have a bad issue with voltage transients and draw an unusually high current. Spikes or dips on the 5v line from regulators not responding quickly enough to robot conditions can cause them to reset, which causes the bot to stop for a few seconds. I noticed this behavior in TB alot, so decided to try it. I approve. It was better that it did this in testing and on the Media Lab floor than in a match against something big and pointy.

At weigh-in, TB came in at 12.0 pounds even. However, the scale measures in half-pound increments, and I haven’t actually put it on a ounce or gram scale yet. So it could be anywhere from 12.00 to 12.24 pounds, which is over. Why the hell is my UHMW box so heavy?

The bot handled great during a quick test drive. I was able to do a fair share of drifting around the test box, and it’s zippy even on 11.1 volts.  Due to the Spektrum capacitor hack, it did not drop out at all.

Anyways, the tournament starts tomorrow. Reports later!

TB4.5MCESP1 Update 17: Last build report!

Feb 15, 2008 in Bots, Project Build Reports, Test Bot 4.5 SP1

After a 14 hour nonstop work session at the Media Lab, SP1 is finally in some form of running order, sans arm ESC of course. I think I was in the building from 5 in the afternoon, through the night, to around 7AM when I decided to stop and take care of the little details at the event.

Usually, in the mad rush to finish, I don’t get around to taking pictures, but I did this time.

Poseur-finish picture with bent wedges. I wandered over to Central Machine Shop and got these made on a shitnormously huge metal brake. Best part? It was gratis. So, big thanks to the guys at CMS, because otherwise I would have to run non-curvy wedges.

One of the fixes I came up with to deter arm jamming was this additional half-bearing/chassis extension which cradles the motor shaft. It actually took three tries to get this right because it was about 3AM and I kept missing at least one small detail which made things not fit.

In the end, it doesn’t prevent the jamming as well as I’d like, because the motor still exerts enough brute force to bend the UHMW. If this piece were aluminum, it would work significantly better.

The last ditch solution is to glue the bottom of the motor to the baseplate such that it cannot flex, at least as much. I might also take the hard stop pins out to give me some more reaction time.

Unfortunately, the loss of that extra flex may have other consequences. I managed to explode a tooth off the pinion during testing. I’m beginning to have a feeling that the lifter is going to be less reliable than I expected.  And I didn’t expect much to begin with.

To fix the offset hinge holes, I popped them on the Media Lab mill and slotted the holes with a 3/16″ endmill. This not only lets me line the wedges back up, but also allows them to be mildly adjustable so I can trim them to the floor.

Speaking of trimming the wedges, here’s one of them in an angle vise. This was just one of about 4 different setups I tried for milling down the wedges, so some of the surfaces are a bit inconsistent. However, since they’re just going to get banged up anyway, it doesn’t really matter.

One finished wedge, with a partially finished one behind it. The mill’s Y-axis travel is too small for me to do both wedges in the same vise setup.

Test mount and gratuitous botshot. Things do fit.
And this is just a gratuitous botshot!

So, as of Thursday morning, the bot does move, and is decently maneuverable. It’s not fast on 11.1 volts, but I’ll manage. I still need to scrounge an arm controller at the event, and take care of some minor issues such as the charging lead and a power light of some sort. I’m not too worried about the charging lead since I can take the top off the bot to charge it.

Oh, and I wrote this from Harrisburg, PA. Event begins tomorrow!