Motor / gearbox Design for 100 lb AGV

Look at the parts sold for the combat bots. They are around the size and power you are looking at. Here's a store that sells to that market with an example high power motor:

I don't know how to estimate how much power you need to push your machine. Is it going to be running on grass or a hard surface? It will need more power to run and turn on grass. Motors and power controllers for these size motors will not be cheap if you want real control over the machine (like R/C control with variable speed and steering). The above store might be able to help you pick parts to meet your needs.

You will probably save money by using 2 motors and 2 speed controllers instead of 4 smaller motors.

1. What's a "shirt shooting vehicle"?
2. How does this thing steer? With four wheels, differential drive won't work very well.
3. See "
   formatting link
   " for motor controllers in the indicated range.

John Nagle Animats

It is a vehicle with an air powered cannon to shoot T-shirts into the crowd of a sports game (basketball) for example. Application is smooth floor. You are correct, manuverability will be important to help aim the shirt. The cannon mount is fixed, and does not spin, the vehicle will need to turn to aim the shirt. Thank you.

I'm also working on a large robot. I'm using two 24VDC wheelchair gearmotors (ebay) and two rubber-wheeled swivel casters (Harbor Freight). I'm not sure why you're using 8 wheels for an application on a smooth floor. This will only serve to reduce maneuverability. Robots with more than 4 wheels are usually used for very rough terrain.

BRW

OK. Go with two powered wheels, two casters, an AX2550 Roboteq controller, an R/C receiver, and some surplus DC gearmotors like these: "

". John Nagle

This is similar in weight and function of a "battle bot," and there are plenty of plans and descriptions around the Web you can follow if you don't want to get into the formulas and equations. "Cost effective" is relative, but the beefier motors and drive controllers aren't exactly cheap. Reduce cost and weight by using two motors and two support casters. If it's good enough to cart grandpa around in his Hover-round, it's good enough for a robot.

-- Gordon

OK, heres the scoop, this is my dads post, and my project....I have a little more info...

we are CURRENTLY using 4 of these motors...

and this is our motor controller (the AX3500)

2 8 in rubber wheels for the back, and 2 8 in omni wheels for the front (allow to be slid both ways...)

2 12V small car batteries, and a 6 channel R\C controller\reciever

The tank shoots t-shirts at MSU sporting events during half time\other events, and hopefully lions\pistons games...

We need this thing to turn\manuver better...The motors we bought were crappy and the gear connecting the motor to the gearbox was falling off the shaft\friction fit...I am not good with electronics, are these motors strong enough and just poorly made? or do we need to look at all new motors\controller?? Any input would be awesome!

I'm sorry, using 8 inch diameter wheels, not 8 wheels. Vehicle has 4 wheels.

This looks like a fine motor for a much lighter robot.

No comment on this because you're using the wrong motors anyway.

First off, you lose torque with wheels this large. You get faster speed, but you need to balance this with the needs of adequate torque. Also, the omni wheels are frightfully inefficient. You're losing a lot right there. Rethink these. They are seldom a practical solution.

Lead acid batteries are extremely inefficient and heavy for their capacity. See if you can upgrade to at least nickel-metal hydride. Li-po batteries would be even better, but very expensive.

Friction fit? The motor shafts appear to use a flatted-D, and so should be secured with a metal set screw. However, for the weight you're talking about, even with lighter batteries, I'd look for something with a key, a hex shaft, or some other configuration that eliminates slip.

-- Gordon

On edit I see you're talking about the internal construction of the motors. Anyway, if this happened to more than one, then I'd suspect the motors simply cannot handle the weight of the robot, and you're demanding too much from them.

I'd do this: get 2 wheelchair motors and two matching wheels. Cost from the typical Web source will be about $300 each set. Sorry, that's just the way it is. Robotmarketplace sells these types of motors, from NPC (a wheelchair parts company). Here is their direct URL:

Supplement these with two swivel casters. You can get heavy-duty ones from the same source as above, and other outlets (Harbor Freight has some nice ones that are larger, and not too expensive).

NiMH battery packs, rated for the voltage/current of the motors you are using. Be sure to get a suitable charger. Remember that many wheelchair motors can be operated at 12-24V, and speed is increased with the higher voltage. Therefore, if you need the speed, double-up the packs to get

24V.

A suitable controller for the motors.

-- Gordon

here is our team website, check out the pics

and for the motors, the gear running from the actual motor to the attached gearbox comes off, not the actual output shaft. it is a tiny gear with no screw on it at all...

We only got those omni wheels because to hopefully make the front wheels slide smoother, we could use orbital wheels maybe? or just get smaller wheels to bring the required torque down...and suggestions on which motors to buy? that would make the 2 wheel drive system work?

URL:

Which motors from that site would you suggest? there are a lot of options...

The best motor you can afford! At $150-300+ each, I imagine the selection will be mainly what kind of money you have to spend.

Chat up the folks at the site and ask their recommendation. They're more familiar with the products, and they may be able to help you select the best motor for the least cost.

As for the other message, I don't see that you need powered omniwheels. Personally I think you're wasting battery power here. A traditional differentially-steered robot with swivel casters (either two on one side, or a set both front and back, depending on design) is tried-and-true.

-- Gordon

As an added note, the NPC site also shows "package deals" that at the least indicates what they think works together. They do show sealed lead-acid batteries, which for the size of these motors may be what you need, but do remember the application they are looking at tends to be combat robots, where the thing needs a HUGE amount of current for a short period of time (usually just minutes).

-- Gordon

Your web site documents say you went with this motor:

Note also the web says they are good for robots in the 10 to 15 lb range. That should have clued you that you were in the wrong range for what you were trying to do.

Very cool.

Looking at the nice write up on motor selection, I see a few issues.

You calculated you needed an RPM of 420, and a torque of 4 ft-lbs (total for all 4 motors combined).

The motor you selected has a no-load RPM of 510 and a stall torque of 1.85 ft-lbs.

Here's the catch which no one told you....

Motors have a torque of 0 at the no load speed, and a speed of 0 at the stall torque rating. These two points can be plotted on a speed vs torque graph and you can draw a straight line between these two points to get a good estimate of what type of torque the motor can produce at different speeds. At half the no-load speed, it will only be producing half the stall-torque. So at 210 rpm, those motors can only produce .925 ft-lbs. That's about half the power you calculated you needed which means the unit will be able to go only half as fast as you wanted - if your torque calculation was correct.

We can use a little math and calculate the torque at your desired 420 RPM and we find this:

1.85 * ((510-420) / 510) = .33 ft-lbs.

So we see at your desired RPM, the motor you bought could only produce .33 ft-lbs of torque or 1.32 total for all 4 which is _way_ short of what you calculated was needed at 4 ft-lbs.

In addition, I wouldn't trust the procedure you used to calculate required torque. You made the calculation based on the rolling friction at the desired speed. However, with hard wheels on a hard surface like concrete or a basketball court, the rolling friction will be insignificant. The power loss due to the transmission between the wheels and the motors (gears? Chain?) and the power loss in the wheel bearings will be determining factors of the torque at speed, not the rolling friction.

However, you used an ungodly high number for the rolling friction coefficient (.1 instead of a more reasonable .01) so who knows if that might produce a reasonable answer or not.

However, even if you had good numbers for power loss in the transmission and bearings, that's not the big load. The big load is what happens when you try to accelerate a bot that heavy from a stand still. The F=MA force is what the torque of the motors have to overcome to accelerate the bot to the desired speed in something less than an hour. :) When a machine that heavy starts from a dead start, or switches from reverse to forward quickly, you will be drawing the full stall current on those motors for a short period.

Then we add to this the fact you are using skid steering to make the bot turn. That will require a ton of extra torque when you try to make it turn which is a complex function of the mass of bot and the friction of those omni-wheels you are using. In addition, only two motors (the non-omni wheels) are required to supply most the torque to make it turn since the motors on the omni-wheels only push it forward and don't help it turn.

All in all, I wouldn't trust your torque calculation. But you might have gotten them in the right ball park anyhow by you extremely conservative guess at the rolling friction coefficient.

However, if you have the bot running with those motors, you can use that get a good estimate of what you really need. How fast does it go when running straight? Can it turn? How fast? How long does it take to accelerate? Using those numbers along with the specs of the motors you have, you can estimate how much more power you need. Tell me how fast it runs with those motors and I'll take a stab a making a guess on what you need.

If it runs at half the speed you want, (and assuming your RPM calculations are correct), we can assume the motor is running at about 210 RPM. Those motors should be putting out about

1.85 * ((510-210) / 510) = 1.05 ft-lbs.

of torque at that speed. The HP produced at that operating point translates to:

hp = torque (in ft-lbs) * rpm / 5252

1.05 * 210 / 5252 = .042 HP per motor

or .17 HP total for all four motors.

In this case, if you want it to go twice as fast, you need around .3 HP total. Or, if you switch to the 2 motor design, .15 hp per motor.

Checking the web site again... This starter kit seems to imply it would work for you sized bot (60 to 120 lb bots):

which uses these motors:

Checking the 24 V chart, we see it puts out .16 HP at 165 RPM and 60.1 in-lbs or 5 ft-lbs. That makes it about 4 times as powerful as your current motors.

So two of these motors operating at that range, should push your bot to around the 10 MPH rating (assuming your current machine is running at about half the speed you want). But it's running at 165 RPM, instead of the 420 you need, so you need to add a 2.54:1 gearing or drive on that to get it into the right speed/torque ratio. Or just use 20" wheels.

But if the current motors can't get your machine running half the speed you want, you will need something larger, or maybe 4 of the NPC-41250s instead of 2.

Or go to an even larger motor like this:

But you will need higher power controllers for that as well. And you are now up to the $330 range instead of the $50 range you started at for each motor. I hope you have a good budget for this project. :)

BTW, don't trust anything I write without double checking it. I have _NO_ experience building these types of bots so I might be completely wrong about something important here. I just enjoy the learning experience of trying to figure out someone else's problems....

Wow, the pictures make all the difference in the world. I see several aspects of your design that make life difficult for your drive wheels.

It looks like you have a set of tank treads on the outside which are mearly cosmetic. However, it also looks like you plan on them working by being dragged along. Consider lightening the load by making the tank treads non-moving and depict them with carved foam and paint.

The width of the wheel base (the real, functional wheels) is fairly narrow when compared to the length of the wheel base. (However, this is only an issue when turning) Consider both the possibility of moving the wheels away from the front and back, and further out towards the sides.

Since you have 4 wheels, you need to realize that you will often be in situations (such as door thresholds and transitioning ramps) where one wheel is up in the air. So, you can end up with one of the wheels carrying half of the weight for a time. You might have been tempted to think that each wheel only carries 1/4 the weight.

You have some hefty pillow blocks there. Perhaps if you went to another wheel support method, you could lighten the load even more.

Another source for affordable wheels and motors would be those made for electric scooters (the toy scooters, not for handicapped) They seem to be under $50 for a motor and wheel. Look at Pep-Boy Autoparts for a store chain that carries them. Here is one source for electric scooter parts;

Another solution may be a hub-motor wheel. Here is one link to a chinese manufacturer. They have some calculations at the bottom of the page to help you choose. Joe Dunfee

parts;

I think the tshirt shooting tank is more like the scooter than the combat bot. If the tank can only do 4 mph and it takes a while to achieve that speed, i dont think its a problem. Maneuverability is more important, like the scooter. Since aiming the tshirt cannon will be done largely by turning the tank, it needs to have a tight turning radius. I like the scooter powertrain analogy. And...isnt everything explained by F=ma? :)

parts;

Yes, the treads are cosmetic. they dont move. Some weight of course but otherwise not a factor. Think the losses in the gear box, and the relatively light weight design of the gearbox is the primary problem right now.

Question: Does the tank get to a usable speed with the parts you are using right now? Or does it strip the gears and stop moving? What is the turning radius with the parts you are using right now?

And btw: the book wikinomics is an excellent read about design collaboration like this group is doing...in your copious spare time, consider reading

When I've seen t-shirts shot into the crowd at a sporting event, the guys doing the work tend to run at a good speed from area to area on the field to try to get to different parts of the crowd and not take up too much time during the game delay. So my guess is that getting the tank out on the court, and moving it from one end of the court to another, is something they can't wait 2 minutes while the tank slowly crawls along at a walking speed. What seems like I good clip when you are standing next to the tank will probably look like it's standing still from from up in the stands in a big arena. If it's that slow, they may have to carry it out, and carry it off the court, instead of letting it drive itself.

Also, with no turret, it can't drive along a side line and shoot into the crowd at the same time, so they have to stop and turn it, shoot a few times in different directions, and then turn and drive it down the court, stop and turn, shoot a few more times etc. All that will waste even more time and bore the fans if the tank isn't able to haul butt. If it had a turret which they could aim and shoot independent of the motion, that would allow them to get away with slower motion because it could be shooting as it moved down the field.