PDA

View Full Version : How to setup a robotic arm (What additional hardware software is needed?)



mr-sk
05-20-2008, 09:15 PM
Hello all,

So, I've got an idea for a robotic arm.

I was looking at the 3x different trossen arms. I'd prefer to get the cheapest one, if it'll do the job.

The website doesn't give much information about how to setup the arm, so I'm looking for some feedback on what is needed once I've acquired the arm.

I want it to interface with a pc..so I'm assuming I can get a USB controller, which I can control via the PC (C, python, whatever). And then hookup that controller to the arm, allowing me to drive the arm from the PC.

My programming skills are ok (I'm a software engineer) but my electronic skills are sub par. I've been learning and building simple circuits lately..but please feel free to go into detail about the electronic/hardware side of this project.

Thanks for your feedback!

- sk

Adrenalynn
05-20-2008, 09:36 PM
>> if it'll do the job

Welcome aboard!

You'll need to define "do the job" first. ;)

mr-sk
05-20-2008, 09:43 PM
Hey Adrenalyn,

Thanks!

Well, the job is to be have the arm controlled by the PC. I'll then be writing some software to run on the pc to automate a process. At first, I'd just want to control the arm manually from the PC. Then I'd move onto writing the software to automate a process.

Is that enough info to get started? I'm concerned about getting the arm connected to the PC, what hardware is required and what programming and hardware resources are available...

heh,

- sk

LinuxGuy
05-20-2008, 09:50 PM
Well, the job is to be have the arm controlled by the PC. I'll then be writing some software to run on the pc to automate a process.
It's important to know what kind of process you want to automate. This could very well affect what arm you need or what parts you need to build it.

8-Dale

mr-sk
05-20-2008, 09:55 PM
Hey RobotGuy,

Well, I had this idea of instead of using the "helping hands" for soldering, I could get a robot arm to hold parts, and adjust them when needed.

At first I'd prefer to get working w/the PC, then I'd want to provide voice activation;

Such as "x 4degrees", which would cause the arm to rotate on the clasped piece 4 degrees on its x-axis. Also, commands like, "open", and "close", etc.

- sk

Adrenalynn
05-20-2008, 10:34 PM
Cute idea!

Consumer servos just aren't that accurate. The new ax12's might be - I haven't played with them. That really dictates where you go as far as hardware is concerned.

Most arms are just a few ounces of holding - under a pound. I can't imagine my frustration at having the thing suddenly take a dive when I put just a few ounces of pressure on a soldering iron.

I think you may be looking at building a stepper-motor controlled arm for that project. I much larger job.

mr-sk
05-21-2008, 07:54 AM
Hey Adrenalynn,

Thanks for the response.

Ok, I didn't realize the servos and holding would be an issue.

"A stepper motor is a brushless (http://en.wikipedia.org/wiki/Brushless_DC_electric_motor), synchronous electric motor (http://en.wikipedia.org/wiki/Electric_motor) that can divide a full rotation into a large number of steps. The motor's position can be controlled precisely, without any feedback mechanism (see open loop (http://en.wikipedia.org/wiki/Open_loop) control). Stepper motors are similar to switched reluctance motors (http://en.wikipedia.org/wiki/Reluctance_motor), which are very large stepping motors with a reduced pole count, and generally are closed-loop commutated."

I see.

So, what's my next step on this project? A rough design, researching parts, etc? I'll keep this thread going as I begin the research.

Feel free to drop links and info that you feel might be helpful.

Thanks again,

- sk

LinuxGuy
05-21-2008, 01:36 PM
So, what's my next step on this project? A rough design, researching parts, etc? I'll keep this thread going as I begin the research.
Design usually happens before you start any project. Defining goal(s), looking at possible ways of accomplishing the goal(s). Researching what has possibly been done before that is similar to what you want to do is also important. You need to learn a lot about these things before you can proceed on a project. Google works well. :happy:

8-Dale

Adrenalynn
05-21-2008, 01:57 PM
I concur that the design on this is most important. It looks fairly unique outside of industrial robots, so you might want to try a design scaling-back the industrial.

I was also thinking multi-segment arm. In this case, you don't necessarily need a lot of DOF. Maybe a pair of AX12 for the shoulder, and then straight to wrist and hand with an AX12 each?

I've been thinking about tendons for awhile. Might have to prototype a tendon system on my crustcrawler arm. Duplicating biology/physiology sometimes makes sense. Millions of years of evolution can't be for naught, right? ;)

metaform3d
05-21-2008, 02:04 PM
Certainly if you're very mechanically inclined then stepper motors might be the way to go. Here's plans (http://www.instructables.com/id/Easy-to-Build-Desk-Top-3-Axis-CNC-Milling-Machine/) for a cheap as dirt 3-axis milling machine that someone built using steppers.

If this is a first attempt at robotics I would probably suggest starting with an arm kit. It's true that it will only hold a few ounces and it might have precision issues, but it does get you up and running quickly and you can spend your time tinkering with it to get it to work better for your application. Servos are mostly standardized so you can often just replace problem servos with ones that are stronger or more precise. or you can add leverage so that a 180 degree servo rotation is only 90 degrees of arm rotation if that's all you need. That increases torque and precision.

What you would need in addition to the arm kit is a controller. If you want PC control probably Phidgets would be easiest. You'd need five channels, so a 4-servo controller and a 1-servo controller together. The whole thing will run $300 to $350, but in my experience this is not a cheap hobby!

mr-sk
05-21-2008, 02:15 PM
@RobotGuy - Of course; I guess I was hoping people will keep chiming in with information. I've had to gear up quickly anytime I take on a new task or hobby. I'm pretty damn good at doing researching, reading and putting it all together.

@Adrenalynn (http://forums.trossenrobotics.com/member.php?u=1711) - Specifically, I'm not familiar with the parts your speaking of, but I'm sure in time I will be.
" Millions of years of evolution can't be for naught, right?" - right on.

I'm readying "Evolutionary Robotics" right now; great book - they're doing some really intresting stuff.

@metaform3d (http://forums.trossenrobotics.com/member.php?u=1705) - Yeah; Getting started with an arm might be the best way to hit the ground running. I have experiance with the phidgets RFID USB reader..works great. I interfaced with it via a c script I wrote, using the phidgets API. I like their stuff alot. So, yeah I'd be open to getting a phidgets controller to hook the arm up too.

Do you have any recomendations (URLs) to arms that you think would be good...antying from Trossen (of the 3 they have listed)? I've been really happy with my orders and customer support from Trossen so far and would continue to provide them business if they have the parts I need.

Thanks for everyones input!

- sk

Adrenalynn
05-21-2008, 02:26 PM
Crustcrawler is an excellent arm (I got mine from Trossen), but he doesn't need all those DOF I suspect, Meta.

How about the Servo Erector Set and then building from there? Short segments = more torque, limiting the DOF and getting rid of the forearm?

mr-sk
05-21-2008, 06:40 PM
Hello,

Thanks - This http://www.lynxmotion.com/Product.aspx?productID=589&CategoryID=108 looks really cool and would let me hit the ground running. I just bought a thinkpad with rs232 and linux to be used as my development box.

I've got OSX and linux VMs, but I find that somewhat of a pain - So I decided to get a new box just for robotics projects.

Anyways, it mentions a windows program, but I'm assuming you could write c to control the arm ...

What do you think about that arm?

- sk

LinuxGuy
05-21-2008, 08:19 PM
How about the Servo Erector Set and then building from there? Short segments = more torque, limiting the DOF and getting rid of the forearm?
I'm just starting out with arms, and in my own usual feet first style, I am jumping in with a 5DOF arm that has one double joint (2DOF) and a 2DOF Little Grip. :veryhappy: Oh, and this arm has another trick - it can fully rotate the gripper/sensor head +/90 degrees (the only positions that really matter). It's going to be a blast learning to control it and writing software for it. :veryhappy: It's 100% SES.

8-Dale

mr-sk
05-22-2008, 08:07 AM
RobotGuy - That are sounds great.

What do you think about the arm above I posted? Think that would work well for version one?

- sk

Adrenalynn
05-22-2008, 10:07 AM
>> Think that would work well for version one

No.

Tear half of it apart and throw it out and it might start to work, but there's no way that gripper will hold your work. That's why I suggested starting with the Servo Erector Set - it's what that arm is created from, and using only one joint in the arm. Even so, that gripper just isn't going to hold your work still.

mr-sk
05-22-2008, 10:27 AM
@Adrenalynn - I was under the impression the 3 arms listed here are part of the Servo Erector Set - http://www.lynxmotion.com/Category.aspx?CategoryID=73 and http://www.lynxmotion.com/Category.aspx?CategoryID=27

Can you provide a link to the item your speaking of?

Thanks!

- sk

Adrenalynn
05-22-2008, 10:34 AM
They're built from the servo erector set. No single link, really. The servo erector set is pieces for building your own stuff. Brackets, metal, etc.

http://www.trossenrobotics.com/store/c/2703-Lynxmotion.aspx

mr-sk
05-22-2008, 11:09 AM
Ah, I see. Very cool. I'm going to draw up some designs this weekend after looking at existing arms and try to come up with a parts lists. I'll post it here and we can move forward!

- sk

Adrenalynn
05-22-2008, 11:27 AM
I think that's a good approach. I'd advise keeping the number of degrees of freedom (DOF) down in this case. The less joints the more holding. A rotating base, a shoulder, and a wrist with 2DOF (pitch and roll).

I have a 6 DOF arm here (not constructed with SES) that I can take closeup photos of wrist and claw and such if you'd find it helpful.

How about using screws instead of servos? The capacity will be hugely higher, and the inherent holding will be infinitely better. I did a screw design for the 'bot we entered in Science Olympiad. That tiny little thing would lift and hold 35lbs with a motor the size of your thumb... Same kinda design you'd find in CAM systems, 3D cutters and printers, etc.

mr-sk
05-22-2008, 01:19 PM
@Adrenalynn - So, instead of using servos and using screws, I'd lose the DOF at that particular "joint"?

The commands I'd like are;

Open - Gripper opens and stays open
Close - Gripper closes on the object (I'm not sure how the sensing for this works ... or if it just closes until the servor/motor cannot "close" any further)
Move[x|y|z - degrees] - I'm not sure if degrees are the best unit of measure. Basically you could tell it to move the piece your attempting to solder so its in the best position relative to you and your soldering iron in your hand.

So the 3 joints you listed, rotating base, shoulder and wrist should be great. I could possibly do away with the rotating base, leaving just the shoulder and wrist; I always solder in the same place, and the arm could be mounted to the surface.

- sk

Adrenalynn
05-22-2008, 01:34 PM
You could do all those with screws. Although the pitch might be a single GOOD servo, like the AX12.

Yeah, continuous rotation micro/small servos to rotate the screws giving you three axis (like a CNC), and the carriage has one AX12 to angle the work. The work carriage could be manually adjusted with a small screw and locking screw so that the work couldn't shift/fall out of the carriage.

So you would have X/Y/Z, [left/right/up/down/in/out] and free rotate towards/away from you for 90deg. The first three axis would have say 30lbs holding, but the rotate would be probably a pound or so total.

LinuxGuy
05-22-2008, 02:34 PM
You could do all those with screws. Although the pitch might be a single GOOD servo, like the AX12.
I would love to design some brackets that would bridge between the Robotis brackets and the Servo Erector Set brackets. I don't think the design would be that hard to do and several different bridge brackets could be designed and made.

I just need good dimensioned drawings for the Robotis servos and brackets. I already have everything I need for the SES side - full dimensions and 3D models I have already created. I'll have to check with Lynxmotion to see how they feel about something like this. I wonder if TR would be interested in this sort of thing..

8-Dale

tom_chang79
05-25-2008, 05:07 AM
My gut feeling is, you're mixing oil and water with that... LM has a relationship with Hitec it seems, and in the area of hobby robotics, Hitec's biggest competitor (in Korea) is Robotis...

If you really want to have a Robotis-based robotic arm, why not a smart-arm made by crustcrawler:

http://www.crustcrawler.com/products/smartarm/images/smartarm1_768.jpg

Adrenalynn
05-25-2008, 10:15 AM
The "why not" is all the pieces available for the SES. What if you want a longer arm? shorter arm? arm with 15 DOF?

My CrustCrawler SG6 is what it is without much expansion (or contraction) opportunity.

Wingzero01w
05-25-2008, 11:26 AM
Dont forget that crustcrawlers making SSB (smart servo brackets) for the AX-12s so they might be compatible with the arms they already sell.

Adrenalynn
05-25-2008, 12:08 PM
Don't forget that the arms they already sell aren't expandable/contractable.

ooops
05-28-2008, 08:19 AM
Just a thought here but I see there are some industrial robotic arms on eBay that are pretty close in price or cheaper than the new hobby arm kits. One in particular uses pneumatic clamps … very high holding power but a new can of worms to actuate it. Would it make sense to “re-program” an already developed arm to do the desired task?
I guess what I am asking is which goal is more important; learning to build the arm, or accomplishing the task?

LinuxGuy
05-28-2008, 08:33 PM
My gut feeling is, you're mixing oil and water with that... LM has a relationship with Hitec it seems, and in the area of hobby robotics, Hitec's biggest competitor (in Korea) is Robotis...
I really don't see where this comes into play at all. If somebody wants to create some brackets that allow connecting the two bracket systems, I see no reason why it can't and shouldn't be done.


If you really want to have a Robotis-based robotic arm, why not a smart-arm made by crustcrawler:
I've just never really cared for CrustCrawler stuff. I don't think it gives me the flexibility I want.

8-Dale

mr-sk
05-29-2008, 08:16 AM
If I had a better understanding of the components involved, I'd be more inclined to piece one together myself.

But since I enjoy programming more, purchasing a ready to go, or an arm that would take minimal alterations to get it working would be good.

As I progress more into engineering and electronics, I'll develop the skills to not have to always purchase a working model; Hopefully, at some point, I'll be able to design, develop and build robotics on my own.

To the point; I'm digging this crustcrawler arm: http://www.crustcrawler.com/products/smartarm/

Its a tad pricey, $800+, but its damn cool looking, hehe.

I've looked on ebay for robotics in the past, but never arms. I'd like to give trossen the business, but the arms they have no one seems to be recommending for this application.

- sk

Adrenalynn
05-29-2008, 10:47 AM
I wouldn't recommend the smartarm either, frankly. I really don't think you're going to find they do the job for the reasons above...