The acrobot is a like a mini-segway – it balances on just two wheels, and can even compensate for pretty big external disturbances (like me poking it in this video). I built the acrobot with my great teammates Joe Polin and Jon Greco.
UPenn’s Mechatronics course, MEAM 410/510, was one of the most intense and rewarding courses of my academic career, and allowed me to create some pretty cool robots over the course of the semester. Mechatronics is the synergistic integration of mechanics, electronics, control theory, and computer science to design and manufacture useful products. It’s an inherently interdisciplinary field, and course instructor Jonathan Feine has done a great job in creating a highly engaging, project-based course with an emphasis on design. The final project, Robockey, is showcased here.
Robockey was the epic capstone to MEAM 510 – a rough-and-tumble five-week project in which groups of three or four students build a team of three little robots that play hockey autonomously – no remote controls, no calling plays from the sidelines, nothing.
The project combines all the key elements of the course: mechanical design, electronics, programming, sleep deprivation, and overcaffeination. My team, comprised of Joe Polin, Jon Greco, Justin Starr, and myself, did quite well in the final tournament, placing third out of seventeen overall. The final tournament was actually a very well-produced public event, and a ton of fun. For more information, feel free to visit the official robockey website.
Here is an awesome video put together by my teammate Joe Polin which showcases our mechanical design and a little game footage:
Our team was red-white-and-blue themed, and our robots were named (sort of) after united states presidents – Millard Killmore, Martin Van Bruisin’, and Grover Cleaveland. I don’t think there was any questioning the fact that we had the most team spirit at the public tournament – a significant portion of Joe and Justin’s fraternity showed up to cheer us on, and we were even featured in the Philadelphia Inquirer.
Some more pictures and videos of our robots can be found below…
Here’s a video of us showing our robots off to a bunch of elementary school kids who happened to be walking by while we were testing. (warning: extremely adorable).
Here are some shot of one of our robots from the front and side. Three circuit boards housed the motor drivers, phototransistor connections, and microprocessor. Gummy silicone tires provided a big traction advantage.
Kaboom! This is what happens when you rapidly thrash positive and negative 12V from a 500mA-rated LiPo through an H-Bridge motor driver. Don’t try this at home, kids.
ASME paid the UPenn ModLab a visit and took some videos of projects going on in the lab. Here you can see Jon Greco and I showing off some modular boat robots for the DARPA TEMP Project (Tactically Expandable Maritime Platform).
I have to admit, I’ve always been a little skeptical of the idea of the “fun-loving tech startup” — companies full of geeks lounging on bean bag chairs and playing ping-pong. The idea that a pampering environment could stimulate creativity and lead to wild success in the great internet frontier always seemed a bit trite. Weren’t these really just a bunch of geeks who got lucky, now attributing their success to some nebulous inspiration found between the primary-colored walls of their offices?
Continue Reading at www.knewton.com →
I had a great time this Saturday at the NYC Music Hackathon, where I hacked, coded, and generally made a lot of very strange sounds for many hours. I teamed up with my fellow Knewton intern Dylan Sherry to hack together some cool signal processing software using Supercollider, an open source programming language for sound synthesis and signal processing.
Dylan and I created software that allows a physical instrument or voice to control the sound of a synthesizer. The program works by processing the input signal to extract the volume envelope and dominant frequency, and then applying these characteristics to a synthesized sound generated with a set of oscillators.
What’s cool about this is that the sound produced has all the harmonic properties of the chosen synth sound (the combination of oscillators determines its frequency spectra), but all the dynamic properties of the original sound (because the original sound determines the volume envelope).
It’s a really cool effect. Dylan and I used a set of pure sine-wave oscillators superimposed at octaves for our base synth, which on its own sounds a little bit like a church organ. When I plugged in my stratocaster, the organ sound suddenly took on its twangy attack, and even became bluesy as I tried some pitch bends. When Dylan plugged in his EWI (a very cool saxophone-like synthesizer), the synth swelled and vibrated through a sustained tremolo note, and danced over the scale through some jazz sax runs.
This is a challenging signal processing problem, and when we set out to tackle it we thought we were going to have to build some really serious tools from scratch. Fortunately, we discovered Supercollider, which has great libraries for extracting features like pitch and amplitude from an input signal. Supercollider is a pretty fantastic (and free!) tool with a good support base, so if you’re interested in playing with sound and computers, definitely check it out.
If we ever get the chance in the future, Dylan and I would like to continue this project and see where else we can take it. Possibilities include extracting dynamic changes in frequency from the input signal (belting while singing, for example, or overblowing a wind instrument) and using them to control the application of a filter to the output signal to give a musician even greater control over the musical quality of the synth sound produced. I personally would also really like to try and package this project into an Ableton plugin, so that I can use it in my own music projects.
Today rounds out my first week at Knewton, where I’ll be interning this summer. Knewton is a medium-sized (~80 people) tech startup in Manhattan that makes adaptive learning software. Knewton’s software uses intelligent algorithms to direct students through modular course curricula which adapts as the student proceeds. The core ideas is to present students with content that will optimally build their knowledge and assessments which optimally gauge proficiency. If you’d like to know more, just visit their website.
It’s a fun place to work. Typical startup atmosphere: very friendly and informal, but hardworking. I get lots of nice perks, too, like free dinner every day.