Advice on motor selection

[andrew888]

Hello all:

I need advice on motor selection for a 150 lb robot. The robot will go about walking speed. No more than 2rps. The motor needs a lot of torque and slow speed. It needs accurate speed control autonomously. It can be run by microcontroller or embedded single board pc. compact size and battery weight will be an issue.

Andrew

[ToyMaker]

Hmmm, walking speed (say 1-3 mph) as less than 2 rps. That says you've already chosen your wheels and done some arithmetic.

For a mobot (mobile robot) in the 150 lb. class the first thing that comes to mind is powered wheel chair motors. Check your yellow pages (paper or electronic) for wheel chair providers. You might find a set for cheap if you do the repair/rebuild (or even free dumpster diving).

High torque/low speed is usually achieved with FET pwm controllers. Two providers are vantec and 4qd; there are many others.

Accurate speed control relies on good feedback. Encoders on the wheels is a common method. This also allows steering by comparing wheel speed.

Compact, light-weight batteries just about mandates Li-ion. One alternative to look at is fuel cells.


robotic regards,

Tom
= = = = =
"Saving Private Ryan. Saving Ryan's Privates."
- - Danny Stern

[andrew888]

Dear Toymaker:

Sorry about me getting back to you late. Your selections are right on. Especially the things you said about accuracy. I would like to use a wheelchair motor.

I plan to run the robot by microcontroller. I plan on using encoders, pwm and pid. I want to make the motor as accurate

as a stepper or servo in positioning accuracy but not holding torque. Do you know if this is possible.

Thank you for the information,

Andrew

[ToyMaker]

Andrew wrote: I want to make the motor as accurate as a stepper or servo in positioning accuracy...

An encoder mounted directly on the motor shaft will allow you to position the motor to a nicety (a 1024 count per revolution encoder results in 360/1024=0.3515625 degree positioning).
The position of your mobot will depend on other factors:
a. tire diameter.
b. play in the mechanical coupling between motor and wheel (backlash).
c. composition of the tire; hard/soft, smooth/rough.
d. composition of the surface under the mobot; gym floor, garage floor, tarmac, grass, sand, etc.
e. distance traveled.
f. number of turns made.
g. and others...
(google "where am I" by Johann Borenstein. He discussed all of this in depth.)

Sorry, I get carried away sometimes.
The short answer is: Yes, you can achieve very close motor positioning accuracy with relatively modest means.


robotic regards,

Tom
= = = = =
"A man can father a baby but a baby fathered does not make a man."
- - Burda Vandeborne

[andrew888]

Dear Tom and all:

I appreciate all your help. I'll always wondered about non-servo/stepper robot drive train accuracy and you have given me the answer.

I need your in-depth answers. I've downloaded the article you listed and now I am studying it.

Thanks again for your knowledge,

Andrew

[ToyMaker]

Andrew:

For accurate position, odometry (wheel encoders or ground sensing) is probably not going to be good enough. I have never seen a mobot that could go out 500 body diameters from a home base and return to within a couple of body diameters using only self-sensing. Some kind of external aid such as a beacon is needed.

So, what is your mobot for? From your original post it looks like you are more interested in speed control than position.



robotic regards,

Tom
= = = = =
"The evolution of computer hardware:
'69 Megadollars for a few bytes.
'96 A few dollars for a Megabyte."
- - Alan Williams

[Adamj12b]

I built a robot about a year and a half ago that weight about 90 lbs right now, using these motors. They have incredible amounts of torque (18Nm Stall). You can get them with long and short shafts. The shafts have bronze bushings in the cases, co i mounted then and then added 1/4" steel plates with bearings to support the shafts to the frame. I am using 8 x 2 inch wheels mounted on custom aluminum hubs. This thing has a load capacity of at least 150 lbs. I can easily rid around on top if it. It can also pull somebody across the floor. I plan on adding encoders to it to know how fast it is going. O, also it is a 4-wheel skid steer drive system with a motor on each wheel. To control the motors I use a Roboteq motor controller. I will add a picture of the bearing mount when i get home. Hope this of some help. -Adam

P.S. When I bought the motors, i called the company and got them for $27 each.

I have attached a picture of the plate that carries the bearing.

[andrew888]

Dear Toymaker and all:

My mobot will be used to help individuals with disabilities to carry items and fetch items. Speed control is important but also being able to start and stop at a certain location on demand, I don't know if that is position control are not.

Andrew

[andrew888]

Dear Adamj12b and all:

Those motors that you are using are awesome, over 13 foot pounds of torque. What speed or rpm do they go.

The price is nice. I will look into them to see if they are suitable for my robot.

Thanks,

Andrew

[Adamj12b]

The robot will go along at a good walking speed. the only problem is that the motors are too powerful to be driving around inside the house. If you accidentally hit something, like a couch or table it WILL move it. I could easialy put 10 or 12 inch wheels on it to make it go faster, the motors have more then enough torque.

[ToyMaker]

Andrew:
About 20 years ago or so a guy named Joe Engelberger developed a hospital mobot called HelpMate. If memory serves it had two cameras, at least two sonars, and a bumper ring in its sensor set. It was one of the first commercial robots. Well worth a web search.
In one of the many papers written about HelpMate it was pointed out that the hardware was the easy part. The really interesting (and hard) part was the software that does sensor fusion, navigation, and safety checks.
Good Luck, and keep us updated on your progress.


robotic regards,

Tom
= = = = =
"Dictionary: a way to airy diction?"
- - Alan Williams