baseline, we tested out the bots with a simple
obstacle course using books of varying thickness.
The hapless Scribbler — not known for its
torque — was hung up on the first obstacle. Our
VEX robot did slightly better, but like many a
frustrated undergrad, it was unable to conquer
calculus on the first try. The VEX bot had trouble
despite being equipped with a gear train built for
torque — it just couldn't get a grip on the tall
With plenty of room for improvement, we
knew that some legs could help the robots have a
better run. However, it would be more than
You Tube videos of the OutRunner from which we
would draw inspiration.
We've always had a fascination with the
challenging mechanics of walking robots. It has
been a topic we've explored dating back to the
very first article of this column, where we
chronicled the trials and travails of making the
Joinmax robotic dog balance on two feet (and a
tail — see the January 2005 issue of SERVO for the
whole story). Our first experience with RHex
motion, however, predates even that.
In the summer of 2004, we participated in an
apprenticeship program put together by the
Institute for Educational Advancement. The specific
program we participated in was held at the Palo
Alto Research Center (PARC) where we were taken
under the wing of a team of engineers led by Dr.
(now Professor) Mark Yim. Specifically, we had the
opportunity to work with modular robotics.
The project we focused on that summer was a modular
snake-like robot designed for tackling tough terrain. The
summer culminated in a search and rescue robotics
competition that was held as part of the Association for the
Advancement of Artificial Intelligence (AAAI) conference in
2004. What makes this experience particularly relevant for
our current project is that the modular snake was not just a
modular snake — like some evolutionary intermediate step,
the snake we worked on had legs.
At the outset of the project, we were given a short
presentation on the RHex robot and the biologically inspired
efficiencies of its motion. We aimed to capture some of
those efficiencies by attaching RHex-like legs to several
segments of our modular snake.
Each module of the snake contained at least two
servos; one would articulate the module around one axis,
While we were designing the robot, we actually tested
numerous leg designs — everything from simple ovals to
ovals with scalloped edges to figure eights. The final design
SERVO 08.2014 71
TROUBLE WITH CALCULUS.
A DRIVING BASE WITH TORQUE.