PDA

View Full Version : Servo requirements



Stobs
10-14-2010, 02:21 PM
One of the projects I'm considering for RG'12 is a 6DoF/leg hexapod, or now possibly a quadrapod instead. Presuming that I go with a 3DoF/leg mech, utilizing AX-12+'s, for the RG'11 event, then I'd add more robust servos as needed for the RG'12 mech. That begs the question as to where the most loading is encountered with leg articulation?

I want to guess that the lowest loading of the leg servos would be on the lateral servo - the servo responsible for rotating the leg back and forth along the mech's longitudinal axis, but logic is telling me that they aught to be stressed relatively equally under load. Yep, basically I'm clueless! :}

I've created a virtual mock-up of the servo positioning, with the servo I'm thinking will be the most lightly loaded highlighted in light green.

[PS: For RG'11 I anticipate going with one gun and as stripped down a mech as I can get away with, and for RG'12 I anticipate a significantly heavier mech, with two guns, expanded sensors and some essentially non-functional eye-candy.]

http://img100.imageshack.us/img100/9058/hexapodservolegloads01.th.jpg (http://img100.imageshack.us/i/hexapodservolegloads01.jpg/)

DresnerRobotics
10-14-2010, 04:00 PM
Can I ask why?

Not trying to be a naysayer, just trying to ring in with reality:

You have 6 months until the competition and zero progress thus far. When I go off on these random project ideas I have to ask myself one question- "What would Fergs do?" He's one of our most experienced engineers involved in this project, and if there is a chance this is something he couldn't feasibly accomplish within this timeline, then there is a good chance your average builder cannot.

You'd need way more than an AX-12+ to simple lift the servo weight, I'd guess RX-64 minimum.

6 dof per leg IK on a hexapod is going to be a nightmare.

Tons of battery consumption with no added payload benefits, if anything, less payload since you're carrying a ton of servos.

No frame of reference to sound off on, unlike the various 3dof walkers that others have built around here.

Advantage to 6dof? Just because it's unique does not make it practical. You will be slower, less payload, have a higher power consumption, exponentially higher price, incredibly more complex system overall, at what benefit?

I can't stress enough how much people underestimate the difficulty of building a not only functional, but competitive (as in, it can hold its own against the now 2+ year veterans), mech. Getting it 100% completed, having time to fine tune everything, and get a ton of pilot practice in is not a quick process at all. I understand the want to 'roll your own' and not stand on the shoulders of those before you, but at a certain point we all need to be realistic about the time constraints and what is actually a practical step. Maybe start with a simple 2 or 3 dof quad/hex and work your way up to something like this? Walk- no crawl, before you try to sprint a marathon.

Stobs
10-14-2010, 04:27 PM
Because, to me, servos are expensive and I have a finite budget. If I could've reused the AX-12+'s from this year's project towards next year's that would obviously be a more economical path for me to take. Since, from your comment, using '12's for such a build would be inadvisable, then I would've considered going to '18's or '24's for this year's mech - accept that you're also recommending a minimum of RX-64's for such a build. So, I now need to consider that the investment in the servos for the relatively small mech I'll attempt for RG'11 will likely have limited reusability for my purposes.

Paul

[PS: The 6th DoF is for a rough concept that I have in mind for a leg's termination; I included the position of all six servos since they are ...they will be the heaviest components and as such were pertinent to my inquiry. Regards]

[PSS: BTW, not to be a nit-picker, but I do take acception, although no offense, at the assumption that I've made "zero progress" - with as much as I have to learn I'm simply in an information-gathering mode, which seems to me to be a bit more prudent that buying a bunch of parts that I have no idea of how they actually go together. Believe me, I do realize I'll have plenty of opportunity to either burn out or assemble the wrong components (and likely both at the same time occasionally!), but I'd like to minimize doing so with at least some "homework" first! :) I'm just trying to ask questions, hopefully moderately intelligent ones, that are helping pinpoint a direction for me to take. Regards]

RobotAtlas
10-14-2010, 05:11 PM
If I could've reused the AX-12+'s from this year's project towards next year's that would obviously be a more economical path for me to take. Since, from your comment, using '12's for such a build would be inadvisable, then I would've considered going to '18's or '24's for this year's mech - accept that you're also recommending a minimum of RX-64's for such a build.

When Tybes said
You'd need way more than an AX-12+ to simple lift the servo weight, I'd guess RX-64 minimum.

I think he meant for the one or two servos in a leg closest to the body.
The remaining 3 (or 4?) servos can still be AX-12+ - that's a fact based on ArbotiX kits we were discussing in another thread.

So if you _really_ want to have 6 DOF (who are we to say not, but I do agree with Tybes about nightmare with 6DOF IK) - you could just start with 3 DOF in 2011 and then next year purchase RX servos and an RX-bridge to drive them and still reuse almost everything from a previous year.
I know several people here went back to 3 DOF from 4 DOF. And Ferg's IssyDoneYet won competition with 2 DOF (quickly reconfigured from 3 DOF, if I remember correctly)

Can you share with us why you need 6 DOF in legs?

Stobs
10-14-2010, 06:18 PM
Thanks for the additional input RobotNV - also much appreciated, along with Ty's. As I mentioned in my initial post, I do plan on going with a limited DoF for RG'11.

lnxfergy
10-14-2010, 08:50 PM
Ok, I've sorta held out saying anything is thread for a bit, but here's my thoughts.

RE: Load on servos. It depends a bit on the stance. If the lower sections of the legs are nearly upright as in your image, the blue servos will carry the most of the load. Of course if they aren't straight up and down, there is going to be quite a bit of load on all the leg servos. As for the lateral servos, they actually will be under quite a bit of load when moving -- in particular, they are most likely to overheat.

RE: 6 DOF. Why? I can see definite advantages with various 4DOF legs: such as abilities to climb, run over taller obstacles, etc. However, going further, I'm not sure you get much gain off 5 or 6DOF compared to the extra weight/power consumption. I know Andrew mentioned IK, but frankly, with 5 servos in a single plain, there isn't much in the way of IK that is going to work on a microcontroller (if you're interested in this, it's typically considered a "tentacle" at that point in the literature I've seen).

RE: IssyDunnYet and 2DOF. This is somewhat irrelevant, there were some issues in how the AX-12 works and what I was trying to do, which led to me running the AX-12s at half power (and thus lower torque). For a modestly loaded AX-12 quad, 3DOF shouldn't be a problem. I'd guess that 4DOF wouldn't cause serious issues either (as long as you scaled back that payload ~220g to offset the extra servos).

-Fergs

DresnerRobotics
10-14-2010, 08:59 PM
When Tybes said

I think he meant for the one or two servos in a leg closest to the body.
The remaining 3 (or 4?) servos can still be AX-12+ - that's a fact based on AirbotiX kits we were discussing in another thread.

So if you _really_ want to have 6 DOF (who are we to say not, but I do agree with Tybes about nightmare with 6DOF IK) - you could just start with 3 DOF in 2011 and then next year purchase RX servos and an RX-bridge to drive them and still reuse almost everything from a previous year.
I know several people here went back to 3 DOF from 4 DOF. And Ferg's IssyDoneYet won competition with 2 DOF (quickly reconfigured from 3 DOF, if I remember correctly)

Can you share with us why you need 6 DOF in legs?


Actually I meant all RX-64s. Using AX-12s as a dexterous 3dof 'toe' at the end of a 3 dof configuration is going to be really sloppy with all that weight on top of them. Even trying to hold position in a vertical alignment is likely going to overload them. RX-64s are heavy. Any time you start mixing servo types, you create more substantial weak points in your limbs with the smaller servos.

Stobs
10-14-2010, 09:52 PM
Thanks for the reply Fergs, much appreciated.


... RE: Load on servos. It depends a bit on the stance. If the lower sections of the legs are nearly upright as in your image, the blue servos will carry the most of the load. Of course if they aren't straight up and down, there is going to be quite a bit of load on all the leg servos. As for the lateral servos, they actually will be under quite a bit of load when moving -- in particular, they are most likely to overheat. ...

Not sure why I phrased that as I did. I realized that unloaded tentacles would be lightly stressed while articulating, and that I didn't know how they'd be stressed in relationship to the rest of the servos while under load, but your reply does answer that point.


... RE: 6 DOF. Why? I can see definite advantages with various 4DOF legs: such as abilities to climb, run over taller obstacles, etc. However, going further, I'm not sure you get much gain off 5 or 6DOF compared to the extra weight/power consumption. I know Andrew mentioned IK, but frankly, with 5 servos in a single plain, there isn't much in the way of IK that is going to work on a microcontroller (if you're interested in this, it's typically considered a "tentacle" at that point in the literature I've seen).

...

-Fergs

The terminating servo won't be involved in the normal gait, and would be utilized under special circumstances - only if I go down this path and I can resolve the mechanical issues I've already been able to visualize with what I'd like to be able to do. Again, if I go this way, with the experience I'll be gaining this year, I'll see what's prudent with the DoF level for next year. I hasten to add that a lot of this endeavor for me is to push - both myself and whatever aspect of the "Envelope" that I may somehow eventually be able to approach, so stopping where most others would as well doesn't hold that much appeal to me. Then again, I don't need to waste time and resources otherwise well spent to re-invent a flat wheel ...after all, I do vaguely recall something about running and crawling someone else mentioned... :)

At this point - as a direct result from the feedback that I've gotten from this thread, the "Rover, 'pod or quad?" thread, and considering my budgetary concerns, I don't see any sense, for me, to maintain a high DoF hexapod on my radar for RG'12. That doesn't mean I won't be undertaking such a project, just that there's no sense in me planning what I'll be doing for RG'11 with an eye towards how I'll transfer resources to RG'12 - which was the underlying reason for posting the thread. Besides, it's possible that I just might learn enough by this time next year to realize that I really ought to consider alternative directions! :)

Paul

Stobs
10-14-2010, 10:08 PM
Thanks for the feedback RobotNV


When Tybes said

I think he meant for the one or two servos in a leg closest to the body.
The remaining 3 (or 4?) servos can still be AX-12+ - that's a fact based on AirbotiX kits we were discussing in another thread.

So if you _really_ want to have 6 DOF (who are we to say not, but I do agree with Tybes about nightmare with 6DOF IK) - you could just start with 3 DOF in 2011 and then next year purchase RX servos and an RX-bridge to drive them and still reuse almost everything from a previous year.
I know several people here went back to 3 DOF from 4 DOF. And Ferg's IssyDoneYet won competition with 2 DOF (quickly reconfigured from 3 DOF, if I remember correctly)

Can you share with us why you need 6 DOF in legs?

I pretty much assumed that TY meant at least from the second (light-blue/aqua colored) servo on down, since all of those would have been involved with articulating either a series of non-loaded servos through to articulating with 1/3 of the load from the terminating servo on back - but good point: I shouldn't have assumed. ;)

I don't know what the RX-Bridge is, and I'm guessing that a discussion for that would be/has been better served under a different thread, but I will certainly look into it - although I'm pretty much set now on forecasting a different set of servo's for RG'12, after we've put the experiences of RG'11 under our belts that is to say.

As I mentioned in my reply to Fergs, the 6th DoF was really just about doing something different and as yet not fully developed. With what I've been beaten up with...uh, I mean, with what I've learned from this thread, if nowhere else, is that accomplishing 'different' isn't necessarily an accomplishment. :}

We'll just have to see what happens, starting with RG'11. :)

Paul

Stobs
10-14-2010, 10:14 PM
... Any time you start mixing servo types, you create more substantial weak points in your limbs with the smaller servos.

Thanks for the info Ty, much appreciated, as that also directly relates to my original inquiry. I wasn't sure if it was worth balancing the cost of a heavier duty servo against trying to refine the stress loading of each servo. The way I was looking at it, in regards to leg articulation, is that since we've got different sized muscles throughout our legs should I design my servo legs in a similar manner. lol, I think that line of questioning has been safely put to bed! :)

Paul

zoomkat
10-15-2010, 12:42 AM
If you want to experiment, get the servo suspected to be the most stressed (probably the light blue one near the shoulder), and attach a yard stick to the servo horn in a horozontal position. Hang weights on the yard stick equal to the servos to be on the arm, and see if you can move the yard stick up and down.