sympathy (yeah, right!).
Here are the gritty details: Hagetaka
was being pushed to the limits on payload
given the wider hip stance. While it
ultimately provides a steadier gait, having a
wider distance between the legs adds to
the leverage required of the hip and ankle
roll servos. The upper torso weighed in at
a whopping 2.1 kg, and as such, I was
playing a dangerous game while creating
the walking gaits of trying to get the
hip-roll and ankle-roll servos with a tight
enough compliance to lift the foot without
dragging.
Removing the battery was an instant
solution, proving I was just slightly over my
payload capabilities. Obviously, the robot
wouldn’t work too well sans battery, and
any smaller battery wouldn’t provide
enough run time for a full 15 minute
match. With only slightly over two weeks
left until competition time, I didn’t think I
had enough time to rebuild the upper torso
using a lighter material, and really none of the equipment
housed could be tossed overboard, so I stubbornly
pushed on.
During my rush to get everything finished, I ended up
damaging four of my RX-64s due to overheating.
Dynamixels have a thermal shutdown which will normally
prevent damage, but I was keeping them just short of that
threshold for extended periods of time, which can be
damaging. Generally speaking, if you’re developing walking
gaits, you want to set this thermal shutdown threshold to a
lower, safer value in the 60oC range. To add insult to injury,
at some point a short occurred which damaged my LiPo
(read: went nuclear) and blew out the Gumstix controller.
The project was more or less grounded; the controller and
damaged electronics would have to be rebuilt, the four
RX-64s would need to be upgraded to EX-106s to handle
the payload with less heat, and ideally, the upper torso
would need a rebuild out of lighter material.
So, I did what any self-respecting tinkerer would do
with two weeks left until said event: I tore everything apart
and started over. Rob Farrell was kind enough to lend a
hand getting my Gumstix controller back online and I set to
work in an attempt to salvage what I had into a
competition-ready robot. About 20 hours of CNC work
later, Giger was born. I fell back to the “tried and true”
humanoid configuration, built a much smaller, more
lightweight upper torso, and changed the battery
configuration into two 1,800 mAh packs mounted
in the feet. Figure 5 shows the fruit of my labor.
Ultimately, my efforts were in vain; I managed to finish
the physical build of the robot the day before I had to leave
for RoboGames and simply did not have enough time onsite
to finalize the walking gaits. This would prove beneficial for
the competition as a whole; leaving me with more time
to focus on running the event itself and making sure
everything went smoothly.
Figure 5
Conclusion
Year One of Mech Warfare went great and given that I
am no longer under a time crunch to get Hagetaka finished, I’ll be able to make the necessary changes to continue the project without having to rush to the finish line.
Giger turned out better than expected, so I’ll likely keep
that project intact for Humanoid Kung Fu competitions and
rebuild Hagetaka with a mix of EX-106s and RX-64s — and
a lighter upper torso. As this project is ultimately funded
out of my own pocket and I’ll be looking at a considerable
cost on the rebuild, the next article will likely be many
months out. I learned a lot through this project, and hope
that what I did right as well as what went wrong is beneficial for all that read this. Keep an eye out for those updates
in the future and until next time, keep building! SV
Gumstix —
www.gumstix.com
Farrell Robotics —
www.farrellrobotics.com
References
Analog Devices ADIS16355 Datasheet —
www.analog.com/en/other/multi-chip
/adis16355/products/ product.html
IP Video Encoder —
www.supercircuits.com
Headplay Personal Cinema —
www.headplay.com
Dimension Engineering —
www.dimensionengineering.com
Trossen Robotics Community —
http://forums.trossenrobotics.com
SERVO 08.2009 51