DPRG List  

[DPRG] segway

Subject: [DPRG] segway
From: David P. Anderson dpa at io.isem.smu.edu
Date: Sun Dec 1 13:02:01 CST 2002


A fellow came by the university last week with a couple of Segway
Human Transport Vehicles and I had a chance to play with one and
ride around a bit, both indoors and outdoors.  

For anyone that's missed it, this is a two-wheeled dynamically balanced
platform very much like my two-wheeled robot nBot.  After playing
with it a bit I have the following observations:

1.  It's a LOT of fun to ride.   The balance is very stable and like
with nBot you make it move simply by leaning fore and aft a small amount.
Steering is controlled with a motorcycle style left-handed grip, counter-
clockwise for right and clockwise for left.  It can spin in place and
we easily maneuvered around the halls in the science building, into
folks offices and around their desks, etc.  

2. Top speed is about 12 mph, which seemed pretty fast indoors and pretty
slow outdoors, easily outpaced by folks on bicycles.  It handled curbs and
ramps with no problem, and had no tendency to roll downhill unless you
expressly leaned in that direction.  Very maneuverable.

3. Turning at any speed is a little counter-intuitive, as it does not
lean into the turn it almost feels like you are leaning out of the turn!
The fix is to lean your body into the turn, like on a four-wheeler or
as racers on motorcycles with sidecars do .  A bit awkward.

On the other hand, you can drive it fore and aft (no turning) without
touching the handlebars at all, just with your feet, with your hands
in your pockets.  It's that stable.

4. Of the two models we had to play with, one would balance by itself
with no rider (though we weren't allow to ride that one) and it had
saddle bags with books and batteries, ballast I suspect.  The other
had no "kick stand" mode and would not balance without the inverted-
pendulum weight of the human.  The one which would balance unattended
was not critically damped, and a small shove left it oscillating for
a long period as it slowly regained equilibrium.  The fellow who brought
them for demo got quite nervous when I started shoving it around.

5. The salesman (he insisted he was not, only "evaluating" them) was a little
defensive on the practical application of these vehicles, and had no good
answers to the questions posed him by the physics and geology faculty (all
of whom had a great time riding!) such as "why this rather than walking?" 
and "why this rather than a bicycle or traditional electric scooter?"  Truth
is for $6000-$9000 I not sure there are good answers to those questions.

(If I was trying to sell them, I would concentrate on the sport market, like
skateboards and surfboards and dirtbikes and sailboats, rather than trying
to convince folks this was a practical replacement for cars or motorcycles
or bicycles or walking).

6. However, that being said, I am more convinced than ever that this is the
OPTIMAL PLATFORM FOR A ROBOT!  In concept at least if not in this particular
incarnation.  I believe our (robot builders') application for this technology
is extremely natural and logical, more so perhaps than for human transport.

7. Most of us are already building three wheel robots, with two drive wheels 
and a tailwheel which is really only there for balance, and causes certain
manuvering problems.  The three wheel design is not really all that stable,
especially on inclines and irregular surfaces, and is fairly miserable outdoors.

Making it more stable involves extended the tailwheel far behind the drive
wheels, and then it tends to run into things when the robot tries to spin in
place.  Four wheels are much more stable than three, and allow sophisticated
suspensions and such not possible with three wheels, but gives up the ability
to spin in place (0 turning radius) without lots of complex technology.  NASA
went to a very complex 6 wheels with rocker boogies and independently steering
wheels on their Mars rover to address these problems.  Quite complex.

8. Military/contruction vehicles solve this problem with tank treads, but these are 
very inefficent, hi-friction and hi-energy consumption.  And because they require
slippage in order to turn, it prevents the use of wheel odometry for location and
navigation, as well as causes problems indoors with loose carpets and marks
on the floor.  Treaded vehicles are also very difficult to maneuver at high speed.
Toy manufacturers in the 90s tried to start a class of racing tanks, but found
the problems of steering at high-speed made it impractical.

For these and other reasons the two-wheel differential drive with free-castering
tailwheel has become the most common robot platform among hobby builders.  The
addition of an inertial measurement sensor and a little software can get rid of the
tail wheel without changing the rest of the design, and _greatly_ increases the
stability, off-road capability, maneuvering, etc.  Hence my increasing confidence
that the two-wheel dynamic balancing technology is a natural platform for robots.

Plus it gets extremely high marks on the Scale of Absolute Coolness.

I shot some photos of the Segway and nBot side-by-side in the hallway of the
Heroy building, for comparison, and put them on


scroll down towards the bottom of the page.


More information about the DPRG mailing list