ABB IRB-2400

Summer 2015-Fall 2017

This arm used to be used for welding and routing in the CMU RI FRC machine shop, but I retrofitted it for experiments in camera calibration. The longest Displayport cable I've ever seen/used connects it to a monitor and wireless keyboard/mouse outside the cage. This was also the only time I ever combined my sewing and robotics hobbies, by sewing some straps to buckle the computer onto the exterior of the arm.

Don't tell anyone, but I actually drove the arm into that table pretty hard once. On accident, of course.

Clearpath Husky (Argus Pattern)

Summer 2014-Spring 2015

39 Playstation Eye cameras arrayed around an otherwise standard Husky with onboard computing (the 40th camera couldn't be mounted due to the rear battery bay). The mass of coiled USB cables made working on this guy tough, but it was worth it just for all the puzzled looks I got.

Portable Flexible Factory Floor (PF330)

Fall 2012-Fall 2014

Design credit: Greg Podnar

A neat circular robot with three omniwheels to keep it nice and stable. I worked on the inter-robot localization system. Cameras detected the AprilTag fiducials on the robot's hexagonal cross-section to initialize laser scan matching of that shape.

H2Bird

Summer 2011-2012

Collaborators: Stan Baek, Cameron Rose, Ron Fearing

I worked on this flapping-winged MAV platform first modified from a commercial toy by Stan, who graduated a short time after I joined. I wanted to make this flier more controllable, so I added a tail prop for yaw and a tail flap controlled by a tiny servo for pitch. I also experimented with some tiny wide-angle optics for an onboard camera. Cameron later extended and used this platform in his thesis work.

In case you're wondering, H2 stands for Humphrey Hu. I didn't pick the name, but sadly it stuck...

Pioneers in Engineering Kit

Summer 2009-2012

Collaborators: Xiao-yu Fu, John Wang, Ryan Julian, and many others...

I helped found a fantastic group called Pioneers in Engineering, which runs a robotics competition for primarily East Bay high schools. The technical core of the program is a flexible, low-cost, easy-to-use kit, composed of mechanical, electrical, and software components all ready for students to use. The first version of the kit was designed by Xiao-yu and Mitch Thompson. John Wang, Ryan Julian, myself, and a few others joined in and helped on the second version. By year 3 we had organized into a sizable organization and somehow I ended up leading the Kit Development team. We ended up with a nice balance of reliable off-the-shelf components and cheap/flexible custom parts.

El Toro XI

2008

Somehow I wasn't fired from engineering team lead in my final year, but I'm sure this last robot made them wish they had. We went with a cable-driven elevator, perhaps as a homage to El Toro 8, but this time with only two rails. Routing the cables under the bottom of the robot was a little dicey, but Ben Steigerwald had the great idea of using bicycle brake sheathing instead of a ton of little pulleys, and they worked great. Having not learned my lesson last time, I proposed again a clever mechanical system to pick up these big yoga balls: Our manipulator had a pair of forks that could rotate up at a 45 degree angle such that pushing the forks down on a ball made them spread out, and then fall back in place underneath the ball's widest part. Then a large pneumatic claw would come down and capture the ball. In practice, however, lining up the robot with the ball accurately enough to get both forks to engage properly was very hard.

Still, this robot was definitely a culmination of our last 3 years of developing and refining MVRT-flavored techniques. Much of the structure was riveted or fastened together with rivet nuts, making it very lightweight. We also made great use of pneumatic compliance for the ball claw. Strict weight management and planning meant that even when clotheslining the elevator on the overpass a few times, this robot was so bottom-heavy I don't think it ever tipped (maybe once when someone pushed us from behind). And of course, we finished slightly underweight in the crate as well. Thank goodness for aluminum bolts.

El Toro X

2007

In my third year on the MVRT I got promoted to the engineering team lead. This year we developed a lightweight riveted outer shell. I also proposed this insanely long arm with a number of crazy features. The main challenge was getting it to fit inside the starting size while minimizing the degrees of freedom we needed. Our solution was a linkage-like system, where only the elbow joint was powered by a very long chain, and another passive chain loop counter-rotated the wrist to simulate a four-bar linkage. At startup the elbow would fold all the way back such that the wrist would also be turned in, but this didn't quite get us in the box, so we had to have an additional base joint that would be deployed on startup by a giant pneumatic cylinder and held in place by a mechanical latch.

I think this robot also ended up slightly underweight in the crate. How we did it I will never know.

El Toro IX

2006

I somehow got promoted to the mechanical team lead in my second year on the Monta Vista Robotics Team, coincidentally coinciding with most of the senior students leaving the team. As such, we had to pioneer lots of unique construction techniques, like large CPVC structures and sliding wheel blocks for chain tensioning. The decision to use 0.5" diameter steel wheel shafts paid off, because we ended up needing to wheelie and slam the robot down to un-jam our ball hopper.

El Toro VIII

2005

One of the first real robots I worked on as a freshman in high school. Patrick Wang mentored Brij Datta and me to design and machine an incredibly overdesigned battery mount. We had to Swiss cheese it later when the robot was overweight, but that's life.