Table of Contents
- Chapter 1: Welcome to Competition Robotics
- Chapter 2: Getting Started
- Chapter 3: Robot locomotion
- Chapter 4: It's All About Power
- Chapter 5: Power Transmission
- Chapter 6: Controlling Your Motion
- Chapter 7: Remotely Controlling Your Robots
- Chapter 8: Strength of Materials
- Chapter 9: Getting to Love the Smell of Victory
- Chapter 10: Brains for Bots
- Chapter 11: The Autonomous: Letting Robots Play by Themselves
- Chapter 12: Robot Sumo
- Chapter 13: Case Study: Building "Live Wires"
- Appendix A: Prototyping Electronics
- Appendix B: Robot References
- Appendix C: Electronic Resources and References
- Appendix D: Mechanical Resources and References
Read an Excerpt
Chapter 3: Robot Locomotion
Moving is what many might call a robot's primary objective; it's
what separates a robot from a plain old computer sitting on the floor. Whether
you use wheels, legs, tank treads, or any other means of locomotion, you've got to
figure out a way for your machine to traverse across the floor or ground, unless
you're trying to build a flying or marine-based machine. The way you make your
robot move will be one of the most important considerations in the design of your
combat robot.
In this chapter, we'll concentrate on locomotion methods that are easy to con-struct
and most effective for large robots and combat machines. We'll also discuss
the drawbacks of some methods for combat robot applications. Several methods
of locomotion have been successfully used in combat and other large robots.
These are legs, tank-type treads, and various other configurations and styles of
wheels. Yes, some really cool machines have used other means to get across the
floor, but "cool" and effective are sometimes very different.
Legs are often one of the first types of locomotion we envision when we think of
robots. For most people, robot means a walking bot like C3P0 in Star Wars or
Robby from Forbidden Planet. However, we must remember that these creatures
were just actors wearing robot suits to make them appear as walking machines.
Walking is actually a difficult task for any creature to perform, whether its human
or humanoid. It takes babies nine months or longer to master the act, and for several
years after that they're tagged with the title of "toddler." A child's brain is constantly
learning and improving this complex process each day. Bipedal (two legs)
walking is really controlled falling—stop in the middle of taking a step and we'd fall
over. Impede the process with a few beers too many, and our built-in accelerometers
(our ears' semi-circular canals) feed us wrong information and we stumble.
Robots with Legs
Watch a person walking and you see them swaying from side to side with each step
to keep balanced. Try race walking and see how exaggerated you must twist your
body to speed up walking. While walking, we always strive to keep our center of
gravity over one foot if even for a fraction of a second. If you count the number of
joints and motions in a person's leg, you'll realize that these joints are multi-axis
joints—not just single-axis joints that we might have in a robot. Many human joints have three degrees of freedom (DOF) in that they can move fore and aft, side
to side, and rotate.
Bipedal robots have been constructed, and a few Japanese companies are demonstrating
these in science news shows. Most robotics experimenters, however,
soon learn the complexities of two-legged robots and quickly move to quadrupeds
(four legs)—and then just as quickly to hexapods (six legs) for their inherent stability.
Sony has sold many of its popular AIBO dogs and cats with four legs, and
the same for the much cheaper i-Cybie, but these machines have many motors for
each leg and are not being attacked by killer robots, as are combat robots.
Hexapods are a popular robot style for robotics experimenters because with six
legs, the robot can keep three feet on the floor at all times, thus presenting a stable
platform that won't tip over. Compare this with a quadruped, which can lift one
leg and easily tip over, depending upon the location of its center of gravity. The
six-legged "hex-walkers," as they are sometimes called, can be programmed to
have their fore and aft legs on one side of the body and the center leg on the opposite
side all raise and take a step forward, while the other three "feet" are on the
floor. In the next step, the other three legs raise and move forward, and so on.
More complex walking motions needed for turning use different leg combinations
selected by an on-board microcontroller. Each leg can use as few as two axes of
motion or two DOF, and some builders have used two model airplane R/C servos
to control all six legs. These types of robots are excellent platforms for experimentation
and for carrying basic sensors, but they are difficult to control and might
present an added complexity for a combat robot's operator.
Although many of the robot organizations you'll find on the Internet focus a lot
of attention on the construction of legged robots, the basic fragile nature of legs
makes them an extra challenge for builders of combat robots. Don't get us
wrong—walking combat robots have been built, and some have done very well in
competition. If you want to build a legged combat robot, go for it. Many popular
robot competitions, including BattleBots and BotBash, even allow an extra
weight advantage for walking bots. Figure 3-1 shows a photo of Mechadon built
by Mark Setrakian. Mechadon weighs in at 480 pounds. This robot is the largest
and most impressive walking robot ever built for any combat robot event. The robot
can roll over, and can crush its opponents between its legs.
If you're a beginning-level robot builder, you'll probably find it easiest to work
with one of the more battle-proven methods of locomotion when designing and
constructing your combat robot. Since we're assuming that a lot of our readers are
still at the beginner level, we'll be focusing on other, less complicated forms of locomotion
for competition robots. If you're interested in learning more about walking
robots, many Web sites and reference books can provide helpful information. Some
of our recommended books and sites are listed in the appendices in this book.
First Person
"I have been
building
mechanical
devices since I was a kid," says
Christian Carlberg, founder and
captain of Team Coolrobots.
Christian is well-known for robot
designs like OverKill, Minion, and
Dreadnought. "Erector Sets,
Lincoln Logs, LEGOs," he adds,
"I used them all."
That early experience with
building toys paid off for Christian,
who further honed his mechanical
skills at Cornell through mechanical
competitions ("build an electric
motor in a couple of hours with
these common house hold items,"
he says). But LEGOs were--and
remain--important. "If you can't
build the premise of your robot with
LEGOs then it's not
simple enough to
withstand the BattleBox."
What competition stands out in
Christian's mind?
"My favorite fight was the
Super Heavyweight rumble for the
first season of Comedy Central's
BattleBots."
Minion's story actually begins
in September of 1999, when
BattleBots announced the new
Super Heavyweight class. "The
idea of building a 325 pound robot
really appealed to me, especially
considering it was a brand new
weight class and there wouldn't
be a lot of competition."
For that event, BattleBots placed
ten 300-pound robots into a box
for five minutes. "I was driving
Minion for that fight," Christian
recalls. "As the fight progressed it
was clear that Minion was the
strongest robot in the BattleBox. I
was pushing three robots at a time,
slamming other robots up against
the wall. It was so much fun and
totally worth all the hours spent on
building the robot."
Indeed, Team Coolrobots exudes
bravado about Minion's power.
"Minion will not break or be
broken. The only way to defeat
Minion is to overpower it. This
used to be impossible but has been
known to happen." Christian admits
that there's a secret to that raw
locomotive power. "The weapon
was always last on my list of
priorities. You can still win as long
as you are moving, which is why the
frame and drive train will always be
a higher priority for me."
Tank Treads: The Power of a Caterpillar Bulldozer in a Robot
Tank treads seem to be the ideal way to make sure your robot has the pushing
power to allow it to decimate an opponent in combat. Hey, they're called "tracks"
because they provide a lot of traction, right?
We'll call the ones robot builders have used "treads" from here on. The military
uses treads in tanks to demolish a much larger and more menacing enemy on a
rugged battlefield. Earth-moving equipment can bounce across rocky ground
pushing many tons of dirt, as the two sets of treads dig in with all their might.
These things seem to be the ultimate means of locomotion for a winning combat
robot. This could well be the situation if the contests were held in a rocky and hilly
locale, but most competitions take place on fairly smooth industrial surfaces. All
the same, let's examine the construction and use of tank-type treads or tracks.
Many first-time robot builders are drawn to treads because they look so menacing.
Treads come in two basic sizes:, massive off-road and toy sizes, and there is no
similarity between the two. The toy variety is just a rubber ring with "teeth" molded
into the rubber. The larger off-road–size treads consist of a series of interconnected
metal plates, "ported by a row of independently sprung idler wheels. The construction
of interconnected plate treads is complex and should be left to experts
with large machine shops. Peter Abrahamson has built a very impressive 305 pound
robot named, Ronin. The aluminum tank treads were custom machined for this robot.
Each side of Ronin can rotate relative to one another thus improving the overall
traction capability of this robot. Figure 3-2 shows a photo of Ronin climbing a log.
Bot experimenters usually opt for the rubber tracks removed from a child's toy
bulldozer, and then start piling batteries, extra motors, sensors, and arms onto the
new machine. When the first test run is started, the rubber tips of the tread surface
begin to bend as they push onto the floor. The robot chugs along just fine until it
has to make a turn. If the operator happens to be monitoring the current drawn by
the drive motors, he'll see a sharp increase as the turn begins. This is one of the major
drawbacks of tank-style treads; they must skid while making a turn, and energy
is wasted in this skid. Only the center points of each "track" are not skidding
in a turn. For this reason, many robotics engineers opt not to use tank-style treads
in their machines.
However, the efficiency of the propulsion system is a less significant factor in
combat robots than in other types of bots. Because a combat robot's "moment of
truth" is limited to a 3 to5-minute match, builders can easily recharge or install
new batteries between matches, making the issue of wasted energy less of a consideration.
With this fact in mind, many builders opt for tank-style treads, so let's examine
another feature of treads: they're complex and hard to mount.
The toy rubber ring tank tread seems anything but complex. It's just a toothy
rubber ring strung between two pulleys. The experimenter with his toy bulldozer
treads might be so preoccupied with the current draw of his drive motors or with
maneuvering the machine that he doesn't notice one of the treads working its way
off the drive spindle. And if the tread slips off your heavyweight bot in a robot
combat match, chances are you'll lose.
Building Tank Treads for a Robot
You've probably realized by now that even the largest toy tracks you can find are
too small for a combat robot or any other type of large robot. The smallest of the
real metal treads are ones you've seen on a garden tractor, and these are too big
for your machine. So if you're dead set on making your robot move with tank
treads, you're probably wondering what to do next. You might start to look at
wide-toothed belts, which work much like the timing belt on your car. The only
trick with using these is that you need to make sure whatever belt you choose has
enough traction to stay competitive on the arena floor. Some successful builders
have used snow-blower tracks, which seem to be just the right size for many types
of combat robots. Flipping a large industrial belt with softer rubber teeth inside
out is another option for builders who want tank treads on their bots. These are
ready-made teeth to dig into the floor, flexible and cheap—what a way to go!
In this case, you go to a friend and have him machine two spindles out of aluminum
that fit the width of the belt. After mounting one of the spindles on a
free-turning shaft and the other to a driven shaft, you try out one of your timing-
belt treads. Almost at once you notice the driving spindle spinning on the
belt's surface when you apply a load to the bottom of the tread. You remember
seeing that the driving spindle on a real tractor has teeth that engage the back of
the tracks. You decide to machine two new drive spindles out of rubber. You're back at your friend's shop and he tells you that he'll have to grind the rubber
down, rather than machine it like metal. After a few hours of experimentation, he
hands you two rubber drive spindles.
Now you have four spindles to mount both belts for a complete robot base, two
rubber and two aluminum. After assembly, you find that the new drive spindles
work pretty well. The rough ground surface of the spindle does a decent job of
gripping the smooth rubber belt's surface. After trying the base out on the floor,
you find that the turning is erratic and decide that you need a row of idler wheels
to keep the entire length of each belt firmly on the floor. Your friend patiently machines
for you 10 idler wheels on which you mount to a series of spring-loaded lever
arms. Wow, this robot is beginning to be a bit complicated! After a few tries on
your garage floor, you begin to notice that the teeth are wearing down. You smile
at your creation and decide to put it away. It was a good learning experience....