by David Geer
Contact the author at geercom@windstream.net
Matt Bunting’s Hexapod
Robot — Takes One and Two
Intel Asks Matt To Build A Second Hexapod Design Just For Them!
It is never too soon to make unique contributions to the field of robotics.
College was a perfect time for Matt Bunting, electrical and computer
engineering student and roboticist at the University of Arizona. Matt
mapped out and built a robust hexapod design that learns to move forward
on its own, among many other interesting tasks and capabilities.
His hexapod earned him the Student of the Year award at the Annual
Creativity in Electronics awards event held in Palo Alto, CA back on May 4th
of this year. Matt also received and fulfilled a request from Intel to build a
similar hexapod to demonstrate its fit-PC2 and Atom processor.
Matt’s first iteration of the hexapod — of which there are now two — was a modestly priced configuration
based on cheaper motors. The first robot used six RX- 28
Dynamixel servo motors, six RX-10s, and nine AX-12s. While
the configuration kept the price of this robot down, it made
it difficult for Matt to readily apply an FTDI-based RS-485
adapter to communicate with the motors as he did on the
second robot. This is because the motor configuration
required half-duplex TTL busses.
The first robot also used foot pressure sensors to
10 SERVO 09.2011
retrieve sensory information from the feet. Matt built a
controller board for this using the same PIC18F4550 he had
used in one of his classes, which came with an onboard
slave USB 1.1 hardware peripheral and code with which to
fully leverage the USB communications.
The second robot — built especially for Intel to demo
what its fit-PC2 and Atom processor could do — used the
FTDI-based RS-485 adapter to communicate with all of the
hexapod’s motors, which include 18 Dynamixel RX-28s used
on the legs and three DX-117s used for the three degree-of-freedom camera operation. The legs were fabricated using
a 3D printer from Stratasys.
The robot’s degrees of freedom are nested in its motor
locations. With 21 servo motors, the robot has 21 degrees
of freedom with each leg having three degrees of freedom.
“One motorized joint permits horizontal positioning of the
foot, and the other two enable vertical positioning of the
foot,” says Bunting. The remaining degrees of freedom are
in the head mechanism, enabling pan, tilt, and twist
motions for the vision camera.
Learning To Move Forward
Matt used a basic reinforcement learning technique
Matt Bunting’s six-legged robot with camera and hardware,
basking in the sunshine.
