ultimately be used as an
educational tool at The Edgerton
Center, the team wanted to
make the scooter’s control strategy as simple as possible so that
one would not necessarily need
an MIT degree to understand it.
With the project’s size and
weight goals in mind, the team
designed a lightweight aluminum base plate, which was
cut with a CNC abrasive water
jet. (MIT has five such machines,
but they are also accessible from
relatively inexpensive online providers, such as www.bi
g-bluesaw.com.) Rather than using the powerful but heavy
24 volt wheelchair motors found on similar projects, the
team chose the inexpensive and lightweight DC motors and
planetary gearboxes featured in the kit of parts provided to
2007 FIRST competitors.
The motors generate a maximum of 343 watts of
power each, giving the scooter a total of just under one
horsepower. The planetary gearboxes provide a 16:1 gear
reduction to increase the torque output to the wheels,
which are 12. 5 inches in diameter with pneumatic tires and
composite hubs. A handlebar was cut
from aluminum extrusion and hinged
to the base plate to allow for
steering control. Figure 1 shows a
CAD illustration of the base plate and
motors, and Figure 2 shows students
assembling these components.
FIGURE 1. Base
plate and motors.
The sensors are small, inexpensive MEMS accelerometers and gyroscopes (angular rate sensors) made by Analog
Devices. One accelerometer senses the angle of gravitational
force, while a gyroscope estimates the angle (with respect
to horizontal) and the angular rate of the base platform. A
second accelerometer detects the angle of the handlebar
for steering. A problem that almost every self-balancing
project write-up mentions is the difficulty of obtaining an
estimate of angle from the noisy accelerometer signal and
the integrated gyroscope signal, which drifts over time.
The team overcame this problem with a very simple digital
FIGURE 2. Assembling the scooter.
Keeping it All Upright
To balance itself, the scooter
implements a feedback control system.
It senses both the angle of the base
platform and its rate of angular
rotation, feeding a combination of
these values back to the motor
controllers to create a corrective
action. For example, if the rider begins
to lean forward, the sensors will detect
the lean and the control system will
send a command to the motors to
drive forward, bringing the base back
under the rider’s center of gravity and
“catching” the fall. The command sent
to the motors is proportional to both
the sensed angle and the angular rate
of the platform, a scheme known as
Proportional + Derivative (PD) control.
SERVO 12.2008 75