robots & me

“Never lose the childlike wonder. It’s just too important.” — Randy Pausch

Early motivation

On a lazy afternoon in the summer of 2010 I was flipping through a magazine when I found an advertisement for a robotics summer camp. For most kids the word robot means a humanoid — intelligent machines that fight monsters, transform into vehicles, and eject missiles. I turned up with those fantasies in mind, and they taught me to build a line follower.

It was the first time I had ever worked with electronics, and I learned a lot, but frankly I was disappointed. The first thing I did when I got home was take the kit apart. I put together some bits and pieces into a torso, a head, and a few makeshift limbs, and screwed them onto my line follower. I still remember the moment the thing moved: I was enormously proud, because I had built my very first “humanoid.” In hindsight the naivety of my 14-year-old self is funny, but it was the pivotal moment that changed my life.

Undergraduate robotics

I spent the next three years assembling random parts into the weirdest robots, mostly without any particular aim. The only thing I knew was that it was fun. Afterwards I chose Electrical and Electronics Engineering, because electronics had been at the centre of most of the robotics I had done and I thought it was the best foundation.

I turned up at my college’s robotics lab on the first day and was turned away — the lab was for graduate students only. But I soon learned the college had student technical teams working on underwater vehicles, solar cars, and even humanoids. Robotics teams are rare in Indian colleges and humanoid teams rarer still, so the chance to work on humanoids felt like destiny. What happened next is on the SRM Team Humanoid page: three 20-DoF humanoids, a lot of falling over, and eventually India’s first RoboGames gold.

The lesson I took from it: if you want to be a roboticist you have to get very, very good at accepting failure. Whatever kind of robot you are building, you are going to fail a lot. The people who become good roboticists are the ones who can accept that, take the lesson, and move on.

Technology that transcends disease and disability

I graduated from SRM University with distinction and joined the LV Prasad Eye Institute’s Center for Innovation as an LVP-MITra Fellow. LVPEI is one of the best eye hospitals in India and collaborated frequently with the MIT Media Lab on new diagnostic technology. I worked mainly on a gaze-tracking algorithm for the Pediatric Perimeter, a device that quantifies the visual field extent of infants.

One day a mother brought her one-year-old to the lab to help us test the algorithm. The child had been declared blind at birth, so we expected no response to the visual stimuli. A few seconds later my algorithm reported that the child’s gaze had traced the exact path of the stimulus. After more testing we worked out that the child was not blind at all, but had a motor disorder. I still cannot forget the mother’s reaction as she held her child.

In the rush of deadlines, projects, and papers it is easy to forget what the work is for. Moments like that one are the reminder. More on this in Pupil+ and eye care assistive technology.

IIT Hyderabad

After the fellowship I came across Dr. R. Prasanth Kumar’s work on extreme legged locomotion at IIT Hyderabad, which lined up closely with my own interests, and joined his lab as a research assistant for a year and a half.

I was floored by the equipment on my first day — enough Dynamixel Pro actuators for two full-sized humanoids, Hokuyo and Velodyne LiDARs, NAO robots, manipulators, and serious machine tools. I felt like a kid in an amusement park. That gave me room to work across bipedal and quadrupedal locomotion, rehabilitation robotics, quadcopters, manipulator control, and computer vision. I also had a long-standing dream of building a full-sized humanoid and making it walk, and I finally did it there — see the IITH full-sized humanoid.

From there I went to the Robotics Institute at Carnegie Mellon for my MS in Robotics, and then to production robotics at Nimble.