PDA

View Full Version : Rover, 'pod or quad?



Stobs
10-12-2010, 05:29 PM
Hi All,

40(+) second version:

I'm looking for suggestions on the "easiest/simplest" mech to build for RG'11. I won't be able to start working on the mech itself, nor begin the requisite programming learning curve, until mid-January '11. I had wanted to start with a fully autonomous caterpillar-based mech, but I'm thinking that the programming overhead is just going to be too much to adequately tackle in the time-frame I have to get the job done, but suggestions on doing so and/or alternative mech types would be greatly appreciated.


For those who'd prefer a bit more background:

I still very much want to get a bot put together for RG'11 but, due to family business concerns, I've been unable as yet to carve the time out to even start - and more than likely I won't be able to get down to starting until this coming mid-January, after I relocate just outside of Tucson, AZ.

Because of this situation I'm rethinking my first bot considerably. I had wanted to start with a caterpillar-based rover that would let me focus on getting a fully autonomous mobile bot into competition this year (RG'11), and then tackle balance, gaits, inertia and whatever [else] would be needed for a quad or biped bot next year (RG'12).

My biggest concern is that it seems to me that there's going to be a lot of programming to get the acquired sensors' data interpreted and processed into output that enables a fully autonomous bot to function "adequately." On the flip-side of that I'm very confident that even if I have to strictly use ready-made components for my first bot (which to me would make it lack any sort of "personality"), that I can assemble a basic bot/mech for RG'11 even in the time-frame that I'm able to allot myself. The crux of the matter is that with the relatively short amount of time I will have to learn the necessary programming from scratch, I'm a bit concerned that I'm going to wind up with the mechanical equivalent of a lump of jell-o instead of a functioning robot if I don't scale back my "vision" appropriately, so suggestions and recommendations would be greatly appreciated.

Regards,
Stobs/Paul

DresnerRobotics
10-12-2010, 05:59 PM
I think jumping into an autonomous mech without building a remote controlled version yet is a big and daunting task for anyone. I highly recommend that you just go with strictly remote control for the first year.

What's your budget looking like? The ArbotiX based quads are easy to build and get up and running, as lnxfergy has done a lot of the ground work with NUKE to get your robot walking remotely. We're even coming out with a kit in the next week that would have everything needed to build a competitive quadropod, you'd just need to add a pan/tilt turret, guns and a camera.

Stobs
10-12-2010, 06:14 PM
Just like most anyone I can find about 12.5 different ways to spend discretionary funds, but I'm thinking I need to put at least $1000 aside, if not $2000, for this if I want to do it right. If that's far short I wouldn't mind hearing what a more appropriate budget would be! :} I know I could get a Bioloid kit for less but, no offense to the creators/providers of Bioloid, building a kit just seriously rubs me the wrong way. Ego or what, I don't know, but I just feel if I'm going to go down that route I might as well as save $975 and get a plastic model kit of a '62 Vette and "enjoy" some glue [which, for the record, isn't my "thing" either]. :robotindifferent:

[PS: That's to say, that if I can get by with under $1000 I'd be darn happy, and I think that's possible, but ...[sighs]...maybe I should just munch my desires down to a 3DoF/leg 'pod or quad and start planning it that way. Just a bit short of what I wanted to do is all.]

DresnerRobotics
10-12-2010, 06:54 PM
Well, look at it this way:

You're probably going to use AX-12s right?

If you have access to machining tools, then by all means build your own brackets. It's pretty time consuming and costly if you don't already have the know how and the equipment though. With the bioloid brackets like $1-2 a piece or $99 for an entire set, it's REALLY hard to justify making your own when timeline is the issue. My advice? Pick up AX-12+'s and some brackets, build a robot, get it 100% operational, and if you have time then go nuts and roll your own brackets and frame. Otherwise you might spend time building all your stuff from scratch, take it 80-90% of the way there, and not be prepared for Mech Warfare. Do not underestimate just how much work has to go into getting functional weapons, video, and control systems in place.

Also, I know you're against kits, but food for though- this kit will be available in a week and is well within your budget:

http://www.trossenrobotics.com/productdocs/assemblyguides/phantomx-quadropod.html

gdubb2
10-12-2010, 06:59 PM
Hey Stobs,

Don't think about a Bioloid kit as a kit.. just a pile of parts. You will need servos, brackets, wires, etc..Set the Bioloid controller on a shelf, get something else, like an arbotiX to run it with. Then just tape a camera and gun on top...

Sounds easier than it is I know, but not so imposing as something autonomous. I've been trying to get a fire fighter off the ground for a year now, and still haven't done it....Yet

Bheka started out as a Robonova-1, but certainly doesn't look like one now. but the Servos, brackets, and other stuff just came in very handy. I even traded off the original controller for some servos. A little aluminum from a scrap yard, and away it went.

Good luck.. now build something..(quads rule)
Gary

Stobs
10-12-2010, 07:48 PM
Thanks Ty & gdubb2, good food for thought. That quad looks good Ty, please let me know when it's available? And that does make sense Gary...just wonder if I'll still be able to look myself in the eyes in the mirror come the morning after I build it! lol :)

Stobs/Paul

gdubb2
10-12-2010, 08:53 PM
I believe Red Green said it best with...
"I'm a man, and I can change...if I have to... I guess"..

G

RobotAtlas
10-13-2010, 02:44 PM
Don't think about a Bioloid kit as a kit.. just a pile of parts.

That was exactly my thought too when I read Stobs' comment about "kit".
Bioloid "kit" is a great time-saver and learning-curve reducer.

Paul, it looks like you need to make a trade-off between the time and how much you want to do by yourself.
If autonomosity is your eventual goal, than you probably want to re-use as much work of others as you can. And that's what Tybes is suggesting too.

Get over the things that many people already spent many-many hours on.
We need you here to work on that autonomous stuff as soon as you can. :)

RobotAtlas
10-13-2010, 02:54 PM
Also, I know you're against kits, but food for though- this kit will be available in a week and is well within your budget:

http://www.trossenrobotics.com/productdocs/assemblyguides/phantomx-quadropod.html


I can't imagine anybody being against this kit. Unless they want a hexapod. Which I would totally understand.

Tybes, these are the best building instructions I've seen in a long while!
I can only imagine how long it took you to do them.

DresnerRobotics
10-13-2010, 10:17 PM
I can't imagine anybody being against this kit. Unless they want a hexapod. Which I would totally understand.

Tybes, these are the best building instructions I've seen in a long while!
I can only imagine how long it took you to do them.

Unless they want a hexapod: http://www.trossenrobotics.com/productdocs/assemblyguides/phantomx-hexapod.html

:D

I have had my personal assistant (Matt Trossen) doing a good amount of the work on these. I just do the mechanical design, overall build outlines, BOM, 3d renders, and keep him from throwing too much feces on the walls.

Stobs
10-13-2010, 10:29 PM
/me nods agreement w/RobotNV, definitely a very nice write up!

Also, I'm wondering what the pro's & con's of quad's vs. hexapods are?

It seems to me you can get away with lighter-duty servos with the hexapod (presuming there's less mass distributed per leg), but that's somewhat offset with having to buy 1/3 more servos - and if the analogous quad could still satisfy design parameters with the same lighter-duty servos it would then be counterproductive? ^.o I'd also think there'd be more gait coding to do with a hexapod?

Stobs/Paul

DresnerRobotics
10-13-2010, 10:44 PM
I forgot to say this: For anyone with a Bioloid kit already, we'll be offering 'upgrade kits' that come with everything needed for these quad/hex bots minus servos.


Gaits are equally difficult I'd say, almost more difficult with a quad at times. Hexapods will have more payload, they're equal in speed to an extent, because once you get up to a certain speed quads have issues staying planted on the ground properly. If you're going the ArbotiX route, Mike has done the vast majority of the hard work for you with the IK and gait engine NUKE produces, leaving you free to focus on things such as control and/or autonomy.

One thing to keep in mind is that for Mech Warfare purposes, with a hexapod you'll be stuck in the Open League (which may or may not have the same density as the Classic league) but with a quad you'd be able to compete in both.

Stobs
10-14-2010, 01:04 PM
I'd have to think that if one wants to advance through a progression of [relatively] more difficult/demanding mechs, that starting off with a quad (rather than a hexapod), would be a pretty efficient way to go.

Let me know if I need to break the following off into a separate thread/heading but, with full autonomy being as difficult to enable as it is, at least at the hobbyist level, then what are the practical qualifications for caterpillar-based mechs in regards to Mech-Warfare?

RobotAtlas
10-14-2010, 01:34 PM
I have had my personal assistant (Matt Trossen)

Tybes, did Matt become your personal assistant after removing your Internet priviledges? :)

RobotAtlas
10-14-2010, 02:14 PM
I'd have to think that if one wants to advance through a progression of [relatively] more difficult/demanding mechs, that starting off with a quad (rather than a hexapod), would be a pretty efficient way to go.

I'm not sure which ones you are saying are simpler: quads or hexes?

According to Mech Warfare, hexes are in the "beginners" league. Which makes sense if you are looking from the point of view of developing a walking gait - quads need to be more dynamic and not as stable as hexes.

This thinking might all change, if/when you start lookiang at hexes as "quads with arms".
Wouldn't you agree that having two extra legs/arms should allow you doing a lot more with your robot?

DresnerRobotics
10-14-2010, 02:21 PM
I'm not sure which ones you are saying are simpler: quads or hexes?

According to Mech Warfare, hexes are in the "beginners" league. Which makes sense if you are looking from the point of view of developing a walking gait - quads need to be more dynamic and not as stable as hexes.

This thinking might all change, if/when you start lookiang at hexes as "quads with arms".
Wouldn't you agree that having two extra legs/arms should allow you doing a lot more with your robot?

Hex's are in the 'Open' League (renamed in updated rules). They're lumped in with tanks and wheeled platforms because all of these platforms present a significant payload advantage. It's mostly a decision based upon payload rather than ease of mobility.

lnxfergy
10-14-2010, 05:13 PM
This thinking might all change, if/when you start lookiang at hexes as "quads with arms". Wouldn't you agree that having two extra legs/arms should allow you doing a lot more with your robot?

I don't think arms matter -- we're mainly concerned here with payload. If it can walk on it's 4 legs it's a quad... if it uses those arms to walk on, it's a hex. Similarly, a 'biped' that walks like a gorilla (see other thread), would be classed as a quad -- since it clearly walks on 4 appendages.

-Fergs

RobotAtlas
10-14-2010, 06:31 PM
If it can walk on it's 4 legs it's a quad... if it uses those arms to walk on, it's a hex.

This gives me an idea. I can take a hex frame and remove middle legs for Mech Warfare.
Then I can compete in both categories. :P

I think I finally decided on a hex myself. Just in time for the release.

Stobs
10-14-2010, 07:09 PM
lol, I used the same discussion to decide upon going with a Quad - funny how people can interpret the same information and come up with completely different positions. :)

gdubb2
10-14-2010, 07:31 PM
Aww.. just fold the 2 middle legs up out of the way when in quad mode.. Of course you are packing extra weight that way...

Gary

lnxfergy
10-14-2010, 08:10 PM
This gives me an idea. I can take a hex frame and remove middle legs for Mech Warfare.
Then I can compete in both categories. :P

I think I finally decided on a hex myself. Just in time for the release.

You may find this isn't optimal. The very long chassis may not walk so well as a quad (especially with a large MW payload).

-Fergs

tician
10-15-2010, 09:46 AM
Not quite the same as an amputee hexapod, but...
http://forums.trossenrobotics.com/picture.php?albumid=87&pictureid=495

I never did test to find its maximum payload before it was twice debrained, but it could carry a mostly full 20 fl. oz. soda bottle (so ~1 pound) without any problems other than payload stability (liquid amplified the rocking motion and the bottle could never sit well in the frame). I would have better tested for the payload if the 8 frame pieces joining the Gerwalks were the F3 (C-frame) instead of the less useful/far less secure F6 (L-frame/toe).

Besides the choice in connecting frames and the lack of an added yaw joint or two, the biggest problem with the entire thing was using RoboPlus and two CM-5s with each independently controlling a Gerwalk. Not only does it love to crash, but the Motion Editor can only handle a single controller at any one time which makes coordinated and fluid motion rather difficult to implement (had to build a data logger to record the poses of one, while using the Motion Editor to program the other).

RobotAtlas
10-15-2010, 05:13 PM
What was the reason for using two CM-5s instead of one?

tician
10-15-2010, 07:52 PM
Partly for the challenge, partly because a CM-5 must be turned on for the bus to be powered (if it has to be on anyway, why not put it to use?), and partly making due with the materials at hand (less than ideal frame pieces, no yaw AX-12+, not wanting to rename all of the Dynamixels on one of the Gerwalks).

This ugly little mutant bird was an assignment: stick two Gerwalks together to mimic the Bigdog and have it walk, trot, run, and climb mini-stairs. Unfortunately the lack of stability/sensors and flimsy joints prevented it from trotting or running, but it could spin in place, walk quickly, and climb stacked plywood stairs pretty well (for whatever that's worth).

jotheberlock
11-01-2010, 11:57 AM
I'm very much looking forward to getting one of the hex kits.

Minor proofreading note on the instructions, though - it's 'Preparation' not 'Preperation' :)

DresnerRobotics
11-08-2010, 12:29 PM
I'm very much looking forward to getting one of the hex kits.

Minor proofreading note on the instructions, though - it's 'Preparation' not 'Preperation' :)

Yeah, Matt doesn't know about spell check apparently. I'm still yet to have read through these before their release.

Kits are coming this week though!