Welcome to A Day in the Life of a Roboticist, our new series featuring robotics professionals from around the world. In each profile, you'll get a glimpse into their daily work with robots, learn about their career paths, and hear their advice for anyone interested in the field of robotics.
To kick off the series, we asked two of Amazon’s leading roboticists to share what it's really like to work with robots every day. In this profile, meet Bhavana Chandrashekhar, Senior Applied Science Manager at Amazon Robotics, who works on developing machine learning systems that help robots perceive and interact with the physical world more intelligently and efficiently.
How did you first get interested in robotics?
I first caught the robotics bug back in undergrad, when a group of us friends built a terrain navigator to compete in an inter-university event. It was a lot of long nights and learning about several areas in a short span of time. As much as I enjoyed demoing it to an awe struck audience, I never thought I would end up working in robotics. I developed my expertise in building software systems and I was always a bit afraid of working with hardware. But when I joined Amazon and was searching for where I could apply cutting edge Machine Learning, warehouse robotics piqued my interest. After my first visit to an Amazon warehouse and seeing the opportunity that lay ahead for robotics, I knew I had to become a roboticist.
What does a day working with robots look like for you?
One of the most exciting thing I get to do everyday is observe the robot in our lab. Each day I spend 20 to 30 minutes just watching the robot, or robots, to take note of what has changed, improved, or regressed. And I try to work out what potential features were enabled on this run — this helps me understand whether the feature we added has truly moved the needle. It is a bit like being a detective.
What are your favorite things about working with robots?
Robots are very transparent — you either made it do something or you did not. You can develop the coolest algorithm but if it doesn't help your robot differentiate things in the wild, you will know almost immediately. The physical rendition of success or failure is very unique with robots and can be very humbling.
What advice would you give someone hoping to start a career in robotics?
Be curious! Robotics is inherently a multi-disciplinary field, so while you can develop expertise in one of those disciplines, knowing about others is also equally important. I have often found the most successful people in robotics are the full stack roboticists who can seamlessly transition between hardware, software, science, and systems. That requires an abundance of curiosity to dwell on how everything works and fits into a robot.
What in your view is the biggest challenge for future robots?
The past decade has seen a transcendent innovation in Machine Learning, especially with generative AI solving problems at super human capability. Translating that to physical world action is hard, and requires closed coordinated motions with hardware. Developing this in a generic way is a big challenge ahead but also a huge opportunity. Imagine if you could build a robot in a matter of hours and you could customize the tasks it needs to do. It would lower the barrier to having robots perform a wide range of tasks in a wide variety of places from homes to industries.
