ITT Enki fails to make robots

[Rococo Modem Basilisk]

Some background: I have a long-standing interest in robots, and I have been trying to build them since maybe age six. I also have a long history of failing to build robots, and my track record hasn't gotten much better -- although I am far more technically proficient than I was when I was six, I don't actually have significantly more funds, and these days people know me well enough to give the disclaimer "don't take it apart" when giving or lending me electronic devices (since I have a tendency to raid them for spare parts).

So, the other day, I was gazing lovingly at the IRC bridge I made for giving one of my bots the capacity for speech synthesis and (crappy) speech recognition, and I thought to myself: "Hey! If I just inserted another filter into this pipe, had a little remote control and a walkie talkie, I could make a little trashcan robot that could zip around and speak gibberish over IRC!". I had recently gotten a first generation roomba from the flea market (with the admonishment: "don't take it apart"), and so I had robots (albiet crappy, insect-like, rodney brooks style robots that really should have had brains built out of discrete components) on the brain.

Well, imagine my surprise when I look at boingboing this morning and find a remote-controlled roomba design with a webcam, etc, built not by taking the roomba apart, but by hooking into some extremely well-hidden nonstandard serial port with a micro!

It turns out, of course, that the roomba model I own is the only one that lacks this serial port, and that unless I break the rule, I will be forced to homebrew some other chassis.

Any suggestions on a chassis I (as a broke student whose three days of spring break are the largest block of contiguous free time he has had since 2009) could reasonably use for a brains-off-board gibberish-spouting trashcan robot? I considered remote control cars, but I'm not sure about how to send back the sensor readings and I'm not confident with soldering up radio transcievers &c.

[Faust]

Some background: I have a long-standing interest in robots, and I have been trying to build them since maybe age six. I also have a long history of failing to build robots, and my track record hasn't gotten much better -- although I am far more technically proficient than I was when I was six, I don't actually have significantly more funds, and these days people know me well enough to give the disclaimer "don't take it apart" when giving or lending me electronic devices (since I have a tendency to raid them for spare parts).

So, the other day, I was gazing lovingly at the IRC bridge I made for giving one of my bots the capacity for speech synthesis and (crappy) speech recognition, and I thought to myself: "Hey! If I just inserted another filter into this pipe, had a little remote control and a walkie talkie, I could make a little trashcan robot that could zip around and speak gibberish over IRC!". I had recently gotten a first generation roomba from the flea market (with the admonishment: "don't take it apart"), and so I had robots (albiet crappy, insect-like, rodney brooks style robots that really should have had brains built out of discrete components) on the brain.

Well, imagine my surprise when I look at boingboing this morning and find a remote-controlled roomba design with a webcam, etc, built not by taking the roomba apart, but by hooking into some extremely well-hidden nonstandard serial port with a micro!

It turns out, of course, that the roomba model I own is the only one that lacks this serial port, and that unless I break the rule, I will be forced to homebrew some other chassis.

Any suggestions on a chassis I (as a broke student whose three days of spring break are the largest block of contiguous free time he has had since 2009) could reasonably use for a brains-off-board gibberish-spouting trashcan robot? I considered remote control cars, but I'm not sure about how to send back the sensor readings and I'm not confident with soldering up radio transcievers &c.

Well if you are worried about  wireless transmission while moving it then you're kind of stuck, not an awful lot of things that would allow you controlled mobility will allow you to jack into them like the roomba.
My advice is either get the roomba with the serial port in it or look at some of the lego mindstorms controlled motors. You can definitely do it with the lego stuff, but it may actually be cheaper to get a roomba, lego isn't cheap any more.
Or be a man and use a remote control car and some bluetooth stuff.

[E.O.T.]

JUST

          glue it all together

ADD

          porn & shellac

LIKE THIS:

         

[Rococo Modem Basilisk]

Well if you are worried about  wireless transmission while moving it then you're kind of stuck, not an awful lot of things that would allow you controlled mobility will allow you to jack into them like the roomba.
My advice is either get the roomba with the serial port in it or look at some of the lego mindstorms controlled motors. You can definitely do it with the lego stuff, but it may actually be cheaper to get a roomba, lego isn't cheap any more.
Or be a man and use a remote control car and some bluetooth stuff.

I was considering disassembling two remote control cars and their respective controllers -- a pair that works on different wavelengths -- and hooking one controller up to a board full of h-bridges to be controlled by the server, then hooking the other controller up to the sensors. Then, I disconnect the actuators for the car attached to the 'sensor' controller, and feed those lines in, maybe using an AVR or something for the D/A lines so I can have serial control rather than using up valuable lines on the parallel port. But, that means disassembling two cars and fucking around with an AVR and a board full of h-bridges (not to mention the fact that I'm not sure whether or not my server has proper raw serial i/o support or if I need to hook in a userspace driver). I would prefer something less wasteful.

It turns out that even the first generation shit roomba was pretty expensive, unfortunately. I don't think there's any hope of me getting another one, second-hand or not.

[PeregrineBF]

What equipment do you have? This matters.

If you have a TIG welder, use Aluminum. It's pretty cheap, and makes for a nice strong chassis.
Steel is rather expensive & heavy, so probably not appropriate.
Brass can be brazed or soldered easily, a benz-o-matic or such works fine. Alloy 260 sheet stock is good, alloy 360 is a bit easier to machine.


The HandyBoard is an excellent microcontroller. It has 4 built in motor controllers for <1.0 amp motors, and a built-in LCD. The BotBoard is similar, but without as many features.

You'd still need motors, etc, but I'm in a bit of a rush, will revise with more info later.

[Rococo Modem Basilisk]

My father has several welders. I very much doubt I will be allowed to use them, nor do I particularly trust myself with them. However, if I am to be homebrewing most of it, I have a more or less complete motorized tanktread-style chassis. Aluminium and steel are overkill, since this is a brains-off-board design -- were I carrying a computer onboard I would need actual pull and stability, but this won't be carrying anything heavier than a cell phone.

I do have an ATMEGA8515, which kind of makes a specialty breadboard-micro-combination unnecessary. Likewise, I'm looking to get an arduino board with all the fixings, but that won't be for several months, if at all. This is just one of the joys of living off pocket change .

I guess ideally I'd want to hack apart a prebuilt chassis and replace the control, doing as little replication of work as possible. This means I'd be best-off if the radio transceivers were prebuilt and I had minimal signals-processing work to fiddle with (say, if I used an existing system for transferring digital data slowly over noisy analog channels, like DTMF tones or something), but what I can't afford or salvage I will need to homebrew.

[PeregrineBF]

Most of my robotics comes from robot-wars style stuff with friends & Botball tournaments (3rd place nationally, yay). The botball stuff is all actual robotics (fully autonomous) whereas this seems more robot-wars style fancy-RC car. I happen to like working with aluminium, it's not that hard to machine. Can be a bit annoying to weld, but nothing too bad once you get used to it.

Stages for designing a bot:
Decide on control scheme (autonomous or remote, or some mix of both).
Decide on drive system (ground/water/air, further subdivided to wheels/treads/walker, jet/propeller/magneto-hydrodynamic, plane/helicopter/rocket, etc etc.)
If it's meant to do things (weapons, lifting, etc) you need servo motors or hydraulics or such. Chose what you need to do.
Select your motors. Do you need servos, how much power used, how much heat generated, etc.
Select your power source. If gas, what engine. If rocket, what fuel mix. If electric, what sort of battery? How long do you need to operate?
How much torque do you need? How many transmission/gearing systems do you have to get/create?
Select your chassis type/material. How strong will it have to be to stand up to the stresses applied?
How will you control your motors? Relays (h-bridge or not?), power FETs, commercial speed controllers?
How many control channels do you need (generally 2 for wheels (1 drive, 1 steers or 1 for each side), 1 for each other servo/hydraulic/etc you need).
You can then determine how complex a radio-control system you need (AM/FM/PM, do you need to take PCM signals, how much interference will there be from/on the power lines to your motors, do you need bi-directional communication, what sort of failsafes do you need in case of interference/controller failure, what frequency band (27MHz, 50MHz, 72MHz, 75MHz, something else) etc, etc.)

[Rococo Modem Basilisk]

this seems more robot-wars style fancy-RC car.

It's fully autonomous. It just has the brains off-board (rather than stick a laptop on the chassis, the laptop is across the room). This is because I don't have the money to buy the parts to build a chassis that can move with eight pounds of laptop on top of it.

[PeregrineBF]

Aaah, ok.... But the motors to move the laptop would (in all likelihood) be cheaper than the 2-way radios, so that's probably not the best choice.

[Rococo Modem Basilisk]

Only if I buy them.

[PeregrineBF]

True. But I can't very well give advice with no real knowledge of what materials you have on hand.

Really, the most basic thing is: what do you want the robot to do? Avoid walls? Follow a line? Collect information about its environment, then discard that info & drive into things?

EDIT: I HIGHLY recommend "Mobile Robots: Inspiration to Implementation" by Joseph L. Jones, Anita M. Flynn, and Bruce A. Seiger. Seiger (now deceased) taught me most of what I know about robotics, it's a very good book for an introduction.