microprocessor installed on
a custom-designed circuit
board. The board was
designed using the PIC
microcontroller-based
educational prototyping kit
and online programming
environment developed
by Machine Science, a
Cambridge-based non-profit
organization that promotes
hands-on engineering
education. Machine
Science’s kit and online
development environment
are easy to use, but
powerful enough to execute
the project’s floating-point
(decimal) filtering and
control calculations. The
main software loop — which reads and filters the sensor
inputs, calculates the corrective action, and generates a
signal for the motor controllers — runs at 100 Hz; fast
enough to stabilize the vehicle’s mechanical dynamics.
Development was also streamlined by a wireless serial
programming and debugging interface that allowed the
team to modify software and view sensor values in real
time with no physical connection to the vehicle. Data can
be viewed and collected remotely from up to 300 feet away
while a rider is cruising on the scooter — a feature that
increases the educational value of the project and which
the team believes is unique to its version. Figure 3 shows
the project’s control electronics.
FIGURE 3. Control circuit board.
filtering technique called a complementary filter, forgoing
more complex and computationally-intensive methods such
as Kalman filtering for an effective and intuitive solution
requiring minimal computation. The details of the filtering
and control algorithm can be found in the project’s full
technical documentation, which is available online (see
Resources).
The control computation is done with an on-board
Finished Product
The project was completed in the fall of 2007, just in
time for its creators to get back to their school work. Since
then, the scooter has made appearances at local high
schools, educational programs at MIT, and other robotics
competitions. It is a bit challenging to learn to ride. Because
the lower power motors require a stiff feedback loop to
provide enough stabilization, it has a tendency to oscillate.
(Wireless data collection reveals the natural oscillation
frequency of new riders to be about 1.3 Hz.) But with a
few minutes of practice, most can get the hang of it and
cruise around fairly well.
More importantly, it draws attention to the design
process, feedback control, and many concepts of
engineering in an inspirational way. All of the hardware is
left uncovered and whenever the designers are showing
off, they attempt to demystify the technology as much as
possible. The full project documentation, including design
notes and source code, is available online. SV
Shane Colton is a senior in the Mechanical Engineering Department at MIT.
He became interested in robotics through his experiences as a FIRST team
member at Plainedge High School in Massapequa, NY. He has been a FIRST
team mentor at MIT for the past four years.
76 SERVO 12.2008
