PDA

View Full Version : [Project] Introducing "Scintilla Maxima"



Stobs
10-15-2010, 03:50 PM
Hi Everyone,

This is an initial posting to see if I'm on the right track or not in getting the build process underway [for my very first bot]. In order to help myself visualize what I need to accomplish to build a bot that's viable for Mech-Warfare, I've broken down an overview of various components/capabilities into what I'm identifying as logical groupings, although this may or may not be how the components are actually arranged during the fabrication process. Tentatively I'm calling "her" Scintilla Maxima, and the following is the breakdown that I've arrived at so far:


Turret
Gun (1)
Amo hopper
M-W target plates
Panning servo
Tilting servo
Targeting - see "Sensor Platform"
Webcam

Sensor Platform
Navigation: no sensors; unit to be remote controlled via webcam.
Target Acquisition: no sensors; target(s) to be acquired via webcam.
Targeting, Final: no sensors; targeting to be fine-tuned with ranging shots.

Carriage
Battery pack(s)
BEC
Communications - XBee/Pro
Controller - Arbotix/Arduino

Locomotion
TBD: caterpillar or low DoF quad

Console
Communications hardware & software
Robotic control programming

I came across a video during yesterday's postings (entitled "My quadroped3", below), and it helped me decide that if I go with a legged bot that I would go with a quadrapod (quadruped?). I am intrigued by the relative simplicity of the locomotive mechanisms of this bot, but I suspect that a translation to a M-W capable bot may not be robust enough - at least not without reworking the configuration to the point that I might as well go with an actual 3DoF quad. I may still decide to go with a caterpillar configuration instead, but at this point I want to concentrate on making sure I understand the overall scheme of the build first, then the "business-end" of the bot (communications, power & munitions), and then I'll decide upon the locomotion.

I think I'm on a proper track here, with what I've outlined above, but I am definitely open to comments and suggestions.

Paul


"My quadroped3"

Stobs
10-24-2010, 10:14 PM
`[Copying a few relevant posts here to separate discussion thread from here (http://forums.trossenrobotics.com/showpost.php?p=43399&postcount=93)...]


Man, that's a great looking mech elaughlin, very cool! I like how you have the camera tucked away into the housing like that, but doesn't that require the quad itself to move in order to perform panning? And did you hack down the Trendnet or go with a different camera setup? Keep up the good work!

Stobs/Paul

[...here (http://forums.trossenrobotics.com/showpost.php?p=43401&postcount=94)...]


Thanks a lot.

The mech does not need to turn in order to pan actually. The pan servo is set into the chassis, but the horn is connected to the floor of the torso. So that whole upper octanganal shaped body spins and the guns and camera tilt.

I am still in the process of deciding between the trendnet and another option. The trendnet camera lens is connected to a large board. So that won't work for my set up, unless anyone know how to disconnect it from the board and use some sort of cables to connect it back to the board.

Otherwise, I am looking into using a small camera lens that can connect to a wifi video encoder.

[...here (http://forums.trossenrobotics.com/showpost.php?p=43406&postcount=95)...]


Thanks for the reply elaughlin, much appreciated.

I'm considering the MS LifeCam (http://www.microsoft.com/hardware/digitalcommunication/ProductDetails.aspx?pid=008&active_tab=systemRequirements) that lnxfergy referrenced, and will very likely get one soon just to see how far down I can hack it. I'm sure I can get the height down since my implementation will provide for the pan and tilt (obviously manual positioning for the product), although I'm not so sure I'd get the cylindrical housing reduced much or even alter the front face significantly at all. I'd have to examine it and the accompanying documentation, but the autofocus feature may well take into account the frontal texture, which would preclude messing much with the facing profile.

Even if I don't touch the housing itself, it's a much more compact unit than the Trendnet but, if I have to start learning ROS, which would necessitate my starting the learning curve on Ubuntu as well, that would put it out of the realm of feasibility for RG'11.

Stobs/Paul

[...here (http://forums.trossenrobotics.com/showpost.php?p=43408&postcount=97)...]


What is your plan? The lifecam is completely different from the Trendnet -- one is a webcam, the other is an internet camera -- i.e. one needs a computer attached, the other just needs an available wireless network to piggy back onto.

-Fergs

[...here (http://forums.trossenrobotics.com/showpost.php?p=43409&postcount=98)...]


It all depends on what I can get done as RG'11 approaches, but off the cuff here's what I'm planning so far:

1) Basic turret fixed to a base for testing with servos driving pan and tilt of one or two guns (was planning on one, but pricing isn't that much for an additional gun and once I've got the assembly worked out for mounting the one then mirroring that for 2nd mounting shouldn't be all that time consuming), along with a hopper(s) as well as the space to mount a Trendnet. If I do go with the Trendnet for this year it will be fixed to the same mounting as the gun(s), so there'll be only one pan/tilt mechanism.

2) Basic carrier to allow for battery position and an Arbotix/XBee setup. Unlikely for this year but I want to allow for using the controller as a spine (might switch to Parallax for that) and going with a brain - i.e., netbook or Fit2PC.

3) Chassis to allow for attachment of the carrier and hardpoints for either legs or cats.

4) 4DoF legs or cats.

That's a pretty rough overview of it but, maybe I should also mention, that I'm considering implementing robotic vision for RG'12 or such, so that's primarily why I'm considering an alternative to the Trendnet.

Stobs/Paul

[...and here (http://forums.trossenrobotics.com/showpost.php?p=43413&postcount=101)...]



Whoops! :o Sorry elaughlin for surreptitiously hijacking the thread like that, mistakenly thought Fergs was pointing that at me - well, it is ALL ABOUT ME, isn't it?! :} :} :}

Regards,
Stops/Paul
No, I was assisting in the hijacking and referring to your comments about the lifecam: because you can't settle on a webcam and then dump the FitPC -- you'll have no video feedback. Honestly, get it walking before bothering with anything else -- if it doesn't walk, it won't matter what else you have working. (further discussion should be moved to a separate thread).

Erik, you can ignore my comments --- they totally don't apply to you (as you already have the trendnet and are showing excellent progress towards RG'11 readiness -- in fact, you're more ready than I am at this point).

-Fergs

Pretty much what I was planning: figured I needed to get the guns [operational], otherwise it wouldn't make much of a Mech, make sure I could initiate control of and communicate with the mech, make sure it could move, and finally see what was happening - hence leaving the space on the turret for the Trendnet, if that's the route I need to go for RG'11 [and it almost certainly will be the case. I'm not building this mech simply for RG'11 as I need to be economical in planning and developing a mech(s) for the next few RG's, building on the help I receive, my experiences and conserving expenses by planning a bit ahead... easily arguably further ahead than I have the knowledge to do so properly, but there in lies the circle of experience accumulation! :} ]

With that[(deleted)], thanks very much for your feedback lnxfergy, it is always very much appreciated.

Stobs/Paul

[some unnumbered PS: I should clarify that I wouldn't be adding the (fit-PC2) initially and would use the Trendnet if I went with a controller (rather than spine-brain arrangement) this year. Quite frankly I'm going to be well pressed - and quietly well pleased - if I simply get a crawling rover with a gun or two and a webcam attached to it for RG'11. I'm very interested in robotic vision though, hence I want to plan for implementing a spine/brain/computer vision setup.]

Stobs
10-25-2010, 10:18 PM
I have a bit more time now than I had anticipated to start working on my mech, and have started compiling a shopping list. With that I need to start getting down to business and I'd greatly appreciate some comments/suggestions on the project time line I'm considering, which I've summarized below. Please keep in mind that I've compartmentalized the design conceptions of the mech into the turret, carrier, chassis and locomotion "assemblies," as mentioned earlier. That's the logical assemblies; physically they may or may not be melded together.

Without working much from experience, obviously, I'm planning for from now through the end of November that I'll be gathering turret, control and communications parts, as well as assembling as much of the turret as possible (since I won't be finalizing a vision system until as late as December).

[(Forgot this part!) December I want to get the power, control and communications parts assembled and confirm their installation - not necessarily that I've gotten them to get the mech to do what I want, but that they have nominal functionality.]

For January I feel I need to work on designing, securing parts for, fabrication of, and assembly of the locomotion system - whether the 4DoF quad I have in mind, or a caterpillar design. All that aside from restarting school that is. :}

For the first two weeks of February I'm planning on ensuring the electro-mechanical completion of the mech, regardless of the final configuration of the mech. Whatever it is I can put together by that time, from a construction standpoint, is what the type of mech it will be; if I have to use a webcam and a remote control tank kit to put a rover into play for RG'11, then I'll have to live with it. I'm pretty confident that I'll know if that'll be the case by mid-January, and have February 1st penned in as the final decision date for the design direction the mech will take.

This will only leave me from mid-February to the first week of April to get the quad walking properly (presuming I get to do the quad build and not the rover), with the last week before the games to tie up lose ends concerning logistics that I may not have covered to that point.

This sounds reasonable from my standpoint, but what the heck do I know?! :} I'm actually quite concerned that I haven't left myself enough time get the software working, and especially gaits.

Regards,
Stobs/Paul

PS: The original concept of "Scintilla Maxima," as had been inspired by the youtube included above, has been shelved. I found the simplistic mechanism for locomotion very appealing, but I don't get the sense that I'll be able to construct a mech based on it that would be maneuverable enough for MechWarfare competition, at least not this year, nor do I think that conceptually speaking it would go in the direction I want to build my experiences on either. I already have a name for my new design, but so as not to spam, I'll continue to utilize this thread unless advised otherwise.

darkback2
10-25-2010, 11:25 PM
Mech Warfare presents a few odd challenges that need to be overcome, and the order in which you do things gives you more or less time depending on the challenge.

Given the plethora of arbotix/AX-12 based bots that will by default be highly mobile and capable, running IK and the like, piloting skills become the major factor. Over the next few years look for people upgrading to AX-18fs if not RX-24Fs. This means that people will be able to move around the field of play really quickly. Its going to come down to piloting skills and reliability issues.

Start out with a design, a clear design for the entire bot. You can't do this piece meal. Trust me, it won't work out in the end, and there will be something that you won't be able to mount somewhere. This will force you to do some crazy adhoc solution that will lead to reliability issues. Make sure you plan out where your targeting plates and scoring system will go. They are both sort of big, and hard to locate after the fact.

You need to be able to move your robot around, and that takes time to learn how to do. Get your body/legs running as quickly as possible. Go with a standard configuration. There is a reason why there is a standard configuration. Get there, and then modify from that point. You can even use a frameset for now, and then make your own leg segments later. That way you won't be under pressure, and you will know that you have something that works.

As for the vision solution, the camera you listed...is it a wifi cam?

Finally,

Squidword is a PC based mech. Its overkill that has to do with my own personal lack of skill in the matter coupled with my desire to make an expensive project multi purpose. It ends up being overweight, and not nearly as maneuverable as all the other bots, and there is very little return on investment. The other thing is that the floor is made from drywall. for a heavy bot that is like trying to walk in mud. Squidwords feet sank in, scaring the floor and rendering the clunky gates I had relatively useless.

I can't really give you any more advice without knowing your skill/experience level.

I hope this helps.

DB

Stobs
10-26-2010, 01:47 PM
Thank you very much for your reply DB, very much appreciated.

I see your points, and they are all very good. I'm trying very much not to incur double expenses, and that's kind of putting me behind the eight ball, if you will. I see very little sense in spending the money for the AX-12+'s this year as I've already determined that to move forward in what I'd like to accomplish servo expenses would be re-incurred next year for RX-24F's - or whatever FotM is most attractive at that point.

I come from a building construction background, but I am technically oriented and have worked as an IMAC (Install-Move-Add-Change) rep responsible for coordinating regional and national software and hardware rollouts for brokerage firms, as well as a broadband support rep for a large regional cable-TV/broadband provider, the latter of which helped me understand the fundamentals of network troubleshooting. I haven't had much hands-on fabrication experience since HS shop classes but the only thing that concerns me with fabricating is getting the right equipment and my eyesight, which is fading a bit as I get older. I had started studies towards getting a certificate in "Advanced Programming" ...bit of a misnomer if you ask me... three years ago, but had to interrupt them a bit over two years ago due to family issues. I'm very much looking forward to restarting my studies this January; with what I've already learned and will be picking up I'm confident that I won't be totally lost with the programming side of things, but I know it'll be far from easy for me, at least for the first few years.

I actually do want to create a stable of mechs - rover, quad and biped, so, putting the unstated but sporadically expressed derisive view of rovers aside (as they pertain to MechWarfare), I'll have to consider putting the legged mech off for this year and do a basic rover for RG'11 and build upon that experience for RG'12 and beyond.

BTW, I think they've switched to a masonite floor for RG'11, so hopefully Squidword will have better footing! :)

Regards,
Stobs/Paul

cire
10-26-2010, 04:49 PM
Well I was going to recommend planning out everything before you start (AKA, plan your base before you build your turret). However, since you are planning on starting with a rover and then going to legs next year, it can make sense to design and build the turret first if you plan on reusing it when you switch from a rover to legs.

If you do plan on reusing it, then I recommend making it as modular and friendly to mount as possible (and light!). Once you build it, it reduces your design options in the future unless you make a new one. Also keep in mind that a rover can hold a lot more payload then a walker can.

For example: You don't want it like my turret, which you cannot take off my bot without taking the whole thing apart, and also it pretty much defines how you have to construct the upper portion of my bot, which would be a lot different if it were a rover instead of a walker.

Stobs
10-26-2010, 05:24 PM
Thanks for your reply cire, very much appreciated.

I really didn't want to start with a rover but, I do want to have a completed project for RG'11, instead of something I bring along in a box that's "95%" complete. :(

Surprisingly (to me), I've had a bit more free time for about the last two weeks, even though I've had to add two more renovation projects to my to do list - which might eat up a lot of my time in the relatively near-term future. Ironically having just submitted a change of address form last Friday with the USPS may have saved my entry for RG'11: I was going to start getting parts for the turret two days ago, but was advised that with the processing of the change of address request I wouldn't get mail for about two weeks. With that I decided to put off doing any online ordering as I didn't want any deliveries getting hung up in shipment and possibly lost. If I had bought the items I was considering I may have, from a budget perspective, inexorably put myself on the path of building a quad or more advanced rover than I have the time and knowledge to complete for RG'11. We'll see how that pans out. :}

On a bit of a 50/50 note, it now seems I may not be relocating back to Arizona early this comming year, as the tenant for my house found a new job; good thing because I really would've hated evicting someone with kids. Serendipitously, I have a lead on a garage that I could take over indefinitely as a workshop just for the price of renovations and labor, which comes about through a long time friend of the family. It's close by to where I'm staying and if we can agree on the work that needs to be done ($$$!) I'd have a halfway decent workshop space even from day one, where as I'd have to build a shop from scratch down in AZ. <shrugs> see how that one goes, but '...when one door closes...' and all that.

Regards,
Stobs/Paul