question about wheels

I don't have any experience with outdoor robots, but many of the ones I've seen are differential drives like what you describe. However, instead of 2 drive wheels and casters, they are either tank like with treads, or have multiple drive wheels, i.e. either 2 or 3 wheels on a side. So, the results are what is called "Skid steering", because when it turns some of the wheels are forced to skid sideways.

The results are that the robot is not as precise when it turns. Also, I think you might not want those black wheels to be skidding sideways on an interior floor because of the possibility of leaving marks.

Joe Dunfee

I am in the process of planning on building an (eventually ) autonomous vehicle for exploring. I have the book Robot Bonanza and I was reading how, on the small robot vehicles, usually what is done is the front wheel is a caster, and the back wheels supply the drive power and steering by rotating one faster then the other. My question is, will this work for an outdoor robot?

I am planning on using a steel wagon for the platform, and I am going to rig it for radio control first, since I have a few usable R/C transmitters and servos. The problem with the wagon is that the wheels that come with it are only 7" diameter by 1.5". In order to move over the terrain I want to, I will probably need 10" wheels and more like 2 to 3 inches wide. I like the idea of steering with the 2 back wheels, but is this practical on an outdoor roving bot? Will the front wheel (or wheels) always recover to straight when the rear motors are run at equal rpm? Or should I plan on somehow rigging a steering motor to the front end too?

Then there's the question of motors, but I just want to deal with one thing at a time since I am in no hurry. I will be using 12 V and then have a circuit using LM 317's to get my 5, 6, and possibly 9 Volts from the battery to drive the servos.

TIA, Joe

Yes, 2 wheels can work. But most "wagons" tend to be long and narrow. For

2 wheel drive, it will work better if it's short and wide with most the weight on the drive wheels. The further the front wheel is away from the back drive wheels, and the more weight it has on it, the more traction and torque the drive wheels need to make the bot turn. If you intend to drive in rough terrain (even something simple like thick grass) the rear wheels might not be able to get enough traction to make it turn. It's just a question of weight and power. You might consider making it run backwards as well - i.e., put the drive wheels in the front and the caster in the back.

Speaking of grass, most of the professional lawn mowers work using the rear wheels to apply power and steer and just use small caster wheels on the front. Here's a picture of one to give you an idea:

As Joe said however, it's more typical when making a bot for rough terrain to use 4 or 6 wheels or to use treads and apply power to all wheels on the same side. ATVs work like that:

As well as skid steer construction equipment like the bobcat loaders:

Their tires skid when they turn so they tend to use wide rounded wheels that can easily skid sideways and they work better on dirt and gravel than they do on pavement because of their need to skid. It's common to turn ATVs into bots because of the simplicity of their drive system. The CajunBot entry into the Dara Grand challenge 2004 for example was built on an ATV:

A small 5 pound model robot skids a lot easier than a large 1500 pound bot of course.

And then there are the bots that skip the caster and only use two wheels:

This is a less than desirable configuration. Anyone who has ever been mud stomping or rock crawling in a 4wd knows that off-road negotiation involves having power to all wheels and a bit of intuition and experience to negotiate obstacles. Teaching computers to do this effectively isn't quite there.

Skid steering is OK but it really needs to involve all wheels that have contact with the ground.

Tracks are good for off-road but there are a couple of caveats:1) they work better on large, heavy vehicles, 2) they are not very power efficient.

Our vehicles are capable of omni-directional movement. This allows a vehicle to crab in any direction and simultaneously rotate around its center. See "ODIS at Superbowl" on a google search for relevant press.

Rob Prowel Staff Program Engineer Kuchera Defense Systems

814-467-9060x232

PS-I have openings for experienced software/firmware, electro-mechanical, or RF engineers who are degreed, US citizens, and can achieve a security clearance. Work would involve living in rural Pennsylvania, midway between Pittsburgh and Penn State .

You need to define vehicle size, wheel diameter, and typical terrain. There are perfectly good power chairs used by the handicapped and elderly outdoors that have a similar design as our typical caster-based robvot -- though it's usually two free-wheeling casters on the front or back, rather than just one. I wouldn't want to recommend using one over rocky or uneven terrain, even without grandma sitting in the thing.

Generally the caster-based robots are best used indoors or outdoors over flat, reasonably level surfaces. Less demanding applications can enjoy the cheapest, easiest designs. Some variation in the surface can be handled by spring-loading the casters, using oversize wheels, etc., but generally another design is just easier all the way around if the terrain is varied. Consider one of the rocker bogey arrangements in use. Look at the Shrimp III for example, which is made for outdoors and can handle some hefty terrain variations. There's also Mars Sojourner Rover that you can study and steal.

-- Gordon

Hello Joe,

Thank you for the reply. Right now, this is still in the concept phase. I did not plan on using the puny little wheels that came with the wagon. I have a set of 10 inch inflatable wheels I salvaged from one of those 2 wheel hand carts (used to move refrigerators and things like that), that are rated at 900 lbs, and have treads like snowblower tires, so they can go thru just about anything. I haven't decided what I will use for the front wheels yet. I apologize for the delay in my response, but earthlink for some reason has had problems with their news server all weekend, and they keep telling me their engineers are "working on it", needless to say, I am about ready to cancel them and go with dsl. But that's another tale for another time.

Skid steering is Ok, this thing is not going to be used indoors anyway, except for on the basement floor which is concrete. My backyard will provide the initial field testing grounds, and then the woods. This is why I wanted to use RC first. Gordon's book, I think it's chapter 8 shows that it is easier to convert to microcontroller from radio control once the bugs have been worked out. Anyway, I will find out once I try it. Thanks again

Joe

Hello Curt,

I had planned on using a deep cycle marine battery for the power supply, and , driving a pickup truck all these years, I understand what you are saying about keeping the weight over the back wheels. If you read my previous post to Joe, you will get some idea of what I am conceptualizing. The wagon platform is only about 10 lbs, but the battery is about 30 lbs, the rest of the electronics I can build, along with the actuators from the RC servos. I am estimating total weight at about 50lbs. It is a very basic concept, but still in my head as I am still getting ready for final exams that will go from the end of this week up until 20 Dec. After that, I will have lots of time to devote to this project. I know snow mobiles work with treads too. I have ruled out nothiing yet. As I said, I am still conceptualizing. There is nothing on paper because I have not really had time to do anything but study and read Robot Bonanza. I am up to the chapter on microcontrollers now, but that's out in the future. I just got a 40% off discount coupon on any book from the book store that I have a discount card with, and last time I was in there, I looked in a book that was about outdoor bots, and wanted that one too, but budget concerns prevented me from buying it at that time. The discount is only for one day, and that day is tomorrow. I will be outside their door at

9AM waiting for the doors to open so I can buy it. I don't remember the authors. It is in the faq for this group. This wagon is one of those radio flyers, meant to hold one kid (probably 65-80 lbs, with the puny wheels that they supply with it, but I won't be using those), My main concern in my original post was, should the two wheel drive in the back and 'caster type' in the front work for this application. I guess I am going to find out. If not, I will investigate the options you cited above. Oh, and thank you for the links. I will check them out. And, I apologize for the delay in the response, but earthlink is having trouble with their news server, so I signed up for google groups which is how I am responding now. Thank you again for the info.

Joe

As the owner of both a six wheel ATV and a zero turn riding mower, I have found the following;

The zero turn mower, which has individual power to each of the rear wheels and two free rotating caster wheels in the front, does not have any control trying to go across a slope. You can go up or down a hill, but not very well accross it because the front is hanging way out there, even though the frame is probably a square dimension. The uphill rear wheel losses traction because all the weight is on the downhill tires, only one of which has any control on steering.

The six wheel ATV with power to all six wheels is, of course, skid steer.

The person recommending it be short coupled front wheels to rear wheels to ease the steering is correct. The other factor is the weight to tire surface area ratio. A 1500 lb machine turns as easily, relative to the power needed to move the machines, as a 100 lb machine if the weight to tire contact surface area stays the same. (ie. the coefficient of static friction between surfaces)

The ATV designer's plan was a good one, to use as much tire surface area as possible. As long as power is going to every surface that is in contact with the ground, you can really oversize the tires and reduce the tire scrubbing when turning.

The ATV has six very wide heavily patterned tires, but barely leaves any traces of traversing over a soft surface. I am building removeable 18 inch wide tracks that fit over the tires for snow. These are offered by the "Mac ATV" manufacturer as well, but not as wide as I want them.

You have to balance the final weight to tire surface area ratio with the type of surface you wish to traverse. The extremes are "floating over" the surface versus "digging into" the ground surface. Everything is a compromise.

Good Luck!