This covers the construction of the motorized base. This includes the circular platform, the wheel/motor assemblies, and the front and rear casters.
Note: Almost all photos are thumbnails and can be seen at full size by clicking on them.
(note: all these requirements are negotiable and your resulting robot will, no doubt, be a compromise)
The base will be a motorized platform to move the "rest" of the robot around.
The base should be able to support and propel a robot of at least 75 lbs (hopefully, never to be exceeded).
The base should be capable of pivoting in place to accommodate simple existing navigation techniques.
Top speed should be around 1.5 feet per second (500 mm/sec). (most Leaf robots are around 3 fps)
The design should be capable of climbing over obstacles of at least 0.5 inches (12mm) in height. e.g. edge of rugs.
The base should provide good stability to prevent the robot from falling over.
The basic plan for the base is to be a simple round platform with differential drive and casters, just as is done with many smaller robots including my firefighting bots. This was done for simplicity since the dynamics and control equations have already been developed.
Since the robot will be tall, it will need casters front and rear to prevent tipping over in either direction. In order to be able to climb over small obstacles without "high centering" the drive wheels, the real caster will be spring loaded so that it can travel a short distance up and down. This ensures that the drive wheels will both always keep weight on them.
This is going to be a relatively heavy robot (50 to 75 pounds) and I hate to think of it "accidentally" running into something or someone at full speed. Hence, it is designed to have a maximum speed of only about 1.5 feet/second with 100% PWM. (ended up with 3 fps)
This is a view of the bottom of the completed base. It shows the two wheel/motor assemblies which are identical except for being mirror images. The solid mounted front caster and the spring loaded rear caster are shown. Also shown is a prototype for a bumper with contact switches mounted in it which did not work to my satisfaction and is not seen in later pictures.
(click picture to enlarge it)
side view shows the left wheel and both casters. Note that the platform is
tilted slightly forward sitting on a soft rug surface. This is because
there is no little weight on the wheels and the rear spring loaded caster tends
to lift the drive wheels and press the front caster down. This effect goes
away as the rest of the robot is installed on top and the drive wheels are
pressed down as far as the front caster. (click picture to enlarge
The platform of the base is just a circular piece of 0.5 inch thick particle board. It is 18 inches in diameter. With the future 1 inch thick bumper installed, that will make the total robot diameter 20 inches (51 cm.).
The front caster is a lovely large one which Gordon McComb no longer sells at budgetrobotics.com (I've got to find a new source!.....Just found one! http://www.mcmaster.com/ their part number 2364T2 for $36.68). One disadvantage of a TALL differential steering robot is that resistance to falling over forward (the most likely direction since it will probably be traveling at highest speeds forward) is proportional to the distance between the axle line of the two drive wheels and the center of the front caster. If this distance is small, the robot can tip over forward very easily. I first tried an inexpensive chair caster which rotates only one direction and pivots around an offset vertical axle. (similar to the caster used in the rear and shown later) Since this caster has to be able to pivot 360 degrees, the axle had to be mounted far enough back that the caster would not hit the bumper when the robot is backing up. When mounted far enough back, the center of the wheel was very far back when the robot was moving forward. (compare dimension A to dimension B below)
And I really wanted a large diameter ball on the caster up front (to roll over minor obstacles more easily) which just makes the problem worse. The caster I used has a 4 inch diameter ball which rotates any direction and can be mounted as far forward as possible. The distance between the caster center and the drive wheel axis will not change with the direction of robot motion.
Since I found the front caster AFTER I had built the platform and the main wheels, it was purely by chance that the caster happened to fit underneath pretty well. The stud at the top of the caster extended to be even with the TOP of the platform. So, I made an adapter plate which screwed onto the bottom of the platform, and allowed a nut to hold the caster to be recessed into the surface of the platform. This way it wouldn't be sticking up and getting caught on any furniture.
The rear caster is a simple pivoting caster like I complained about above. But, I think that backing up will always be done at a slower speed and hence the probability of tipping is less. I may be tempted in the future to try a smaller version of the front caster.
As you can probably see from the drawing, The caster is mounted at the end of a 1x2 board which is attached to an inexpensive hinge (Home Depot) which connects it to the base. A spring is made from music wire (hobbypeople.net)
A lanyard is installed to keep the spring from driving the caster all the way down when the robot is lifted. This can be seen in the side view of the platform above. Seems a little kludgy, but it works
(need some specs and dimensions on building this)
Now, the trick part is the wheels and motors. The general requirement is for a setup which has fairly large wheels, for driving over minor obstacles, enough torque to drive up ramps or over those obstacles and encoders on each motor/wheel so that they can be controlled using PID algorithms.
I chose to use some nice Pittman motors which are occasionally available on ebay.com. These are gearmotors with a 5.9 reduction to the output shaft. The motors are rated at 30 vdc, but I found they ran fine at 14.4 volts for my firefighting robots and run well at 12 volts on this robot. I gear them down an additional 6 to 1 using a belt drive system to increase the torque and reduce the wheel speed. I also made adapters to attach the toothed pulleys to the wheel using a lathe which not everyone will find convenient.
My motors were Pittman part numbers: GM9236C534-R2. You can look up the specs on most all Pittman motors at pittmannet.com.
These particular motors may not be available, so I suggest you come up with your own wheel motor setups. It just has to turn wheels of about 6 inches in diameter at about 60 rpm.
The other key item is the encoders. My Pittman motors came with an encoder which provides 512 pulses per motor shaft revolution. This comes out to about 18000 per wheel revolution which is MUCH more than is necessary (giving about 40 clicks per millimeter). Any encoder giving at least 1 click per millimeter should be adequate.
That said, I'll give you some pictures of what I built just for ideas.
This picture (click for enlargement) shows the completed motor and wheel assembly. Everything is mounted to a piece of aluminum angle of non-critical size which has room to mount the motor, and mount the wheel at a distance apart suitable for the belt length. The motor is just attached to the aluminum plate with three screws. In order to provide for belt tensioning, one screw is mounted in a fixed circular hole to allow pivoting the motor while the other two are in slotted holes to permit the pivoting. The wheel is mounted on an axle shaft which is a 1/2 inch bolt (or is it??). The half inch bolt is silver-soldered to a small steel plate to permit it to be firmly bolted to the aluminum.
The wheel is from a caster available from Harbor Freight (link, part number). It has internal bearings and an inner race, hence is fixed tightly to the axle.
Note that you can see my H bridge attached to the aluminum plate in the above picture. I figured it would be well protected being high above the ground and behind the motor. WRONG! Somehow, the LM18201 H bridge chip got bent down flat. So, a little better protection in probably in order.
The axle is shown here. It has a self locking nut (nylon insert) to hold the wheel firmly in place and uses washers as necessary to shim the wheel in position to align the belt on the pulleys.
The picture to the right shows the belt drive. The pulley on the motor end is easy, it just slides onto the motor shaft and locks in place with a setscrew.
(note: if you actually get the motors I have, you will likely have to remove a pulley that comes on it. This can be done by grinding through the aluminum pulley with a Dremel tool, down to the steel motor shaft, and then pulling it off. You'll then find that the shaft has been scored to make a really tight fit for the pulley you just removed. You can remove the scoring to get the new pulley on by running the motor at a moderate speed and sliding a small file back and forth along the shaft.)
The trickier part is getting the pulley attached to the wheel. I did it by making an adapter plate which the pulley fits firmly over and then carefully centering the adapter on the wheel. It is then screwed directly into the plastic around the motor hub which provides a good solid mount. And, you may note that I had to cut out the center of the toothed pulley so that the adapter and pulley would fit around the protruding axle of the wheel.
Note: The Pittman motors are no longer commonly available except at MUCH higher prices. And the belt and pulley system used here is fairly complex to build. Other solutions have been found for a drive system. Just ask on the mail list.
platform. 1/2 inch thick particle board. 18 inches in diameter approx. $ 2
front caster. 4 inch diameter ball caster approx $30
2 inch diameter pivoting furniture
approx $ 4
rear caster arm: 1x2 wood, 1 foot long $ 1
rear caster arm hinge: ??? approx $ 4
rear caster spring: ? dia. music wire hobbypeople.net?? approx $ 2
drive wheel: (2 of) 6
inch diameter harborfreight.com??
approx $ 45
drive motors: (2 of) surplus Pittman approx $ 50
motor pulley: (2 of) sdp-si.com pn??
wheel pulley: (2 of) "
Belt (2 of) " total pulley approx $40
Aluminum angle (2 of) approx $10
Aluminum for wheel/pulley adapter approx $10
Steel plate for axle mount (2 of) approx $ 1
Axle bolts (2 of) approx $ 5
Nuts, bolts, other minor
TOTAL approx $219.