Amazon Robotics AI leaders believe now is a 'particularly good time' to explore careers in robotics
Siddhartha Srinivasa, director of Amazon Robotics AI, and Nia Jetter, Amazon Robotics AI senior principal technologist, discuss inspiration, their roles at Amazon, and tips for pursuing a robotics career.
On October 6, Siddhartha (Sidd) Srinivasa joined Nia Jetter, Amazon Robotics AI senior principal technologist, to discuss the field of robotics, Amazon robotics initiatives, where they get their ideas from, and advice on starting a career in robotics.
Jetter, who earned a bachelor of science degree from MIT in math with computer science, and a master's degree in aeronautical and astronautical engineering from Stanford University, joined Amazon earlier this year. Previously, she spent 20 years in the aerospace Industry, including more than 18 years at Boeing where she rose to become a technical fellow in autonomy and AI.
Srinivasa joined Amazon as director of Robotics AI in 2018, and since 2017 has been the Boeing Endowed Professor at the School of Computer Science and Engineering at the University of Washington. Prior to that, he was the Finmeccanica Associate Professor at the Robotics Institute at Carnegie Mellon University where he founded the Personal Robotics Lab in December 2005. Srinivasa, who describes himself as “a full-stack roboticist, with a focus on robotic manipulation”, has worked in the robotics field since 1999.
Srinivasa, who is an IEEE Fellow, was also a first-wave founder of Berkshire Grey, a robotics company using machine vision and AI to solve material handling problems, and has led Intel’s research in robotics, the Quality of Life Technologies NSF ERC, the DARPA ARM-S, DARPA Robotics Challenge, and the HONDA Curious Minded Machine program. His algorithms have run on the NASA Robonaut and the Mars Rover and he is an editor for The International Journal of Robotics Research.
The entirety of their conversation is above, including why Srinivasa considers himself an “accidental roboticist”, why the “democratization of robotics” is an essential hurdle to clear, and why now is a particularly good time to explore robotics. Below we have excerpted some answers from their wide-ranging conversation on career advice, sources of inspiration, and the field of robotics in general. Editor’s note: Some of these answers have been edited for length.
Advice for those considering robotics and AI careers
Srinivasa: “First, I think you should do it! Stop what you are doing and work on robots! Robotics is still at its infancy. This is both a blessing and a curse, more a blessing. Unlike other fields that require tens of years of work to perfect how to use an instrument, or perfect how to develop techniques, or even learn the language by which you can describe problems and solutions, the textbooks for robotics have yet to be written. There are a few. The barrier for entry into robotics is really low, particularly if you are in adjacent fields.
“Do something that puts you into a state where your work is relevant. If you are undergrad or grad student, I would recommend that you go find an internship in a place where robotics is actually the core business. Not robotics is something cool and fancy to have, but where robotics is actually material to the core business. Go through that experience. Live through that experience, be put through the fire of actually having to deliver something that matters. I think a lot of people who talk a lot about robotics often haven't the experienced the fire of production and delivery, and I think there is a lot of clarity that comes with that.”
Why increasing diversity in robotics and AI matters
Jetter: “I’m hugely passionate about lowering the barrier of entry for understanding topics like artificial intelligence and robotics. I genuinely believe that, through lowering the barrier of entry, that will allow us to increase inclusion and diversity of thought and truly be able to allow us to solve some of the most challenging problems technically, optimally, and most efficiently.”
The modern myth of robotics
Srinivasa: “One thing that we often get misled by is we look at YouTube videos of robots and we think, ‘It’s all solved, everything’s solved, this thing can do a backflip.’ The challenge is that it’s not the one time it does a backflip, it’s the 50 million times it doesn’t and it needs to do. That’s what I find fascinating. It is really about closing the loop and figuring out what to do when things go wrong that is the most critical aspect of robotics. When things go right, the YouTube video is easy to do, but taking something from 80 percent to 96 percent where you are systematically and methodically addressing all the things that go wrong, that's the most important learning for someone to get and I think that's where the real roboticists get their most joy, in taking something from 80 to 95 or 96 percent.”
On choosing to work in robotics at Amazon
Srinivasa: “I was finishing up with Berkshire Gray and I was just being a professor, just happy being a professor, and I did get a call from a bunch of places about joining and being part of their efforts. I asked them all one question: ‘Why robots? Why do you need robots? All I know how to do is build robots and that’s all I want to do, so why do you need robots?’ I found the answers from several of the others very tenuous, which was ‘We want to solve AI’ – whatever that means —'dot dot dot, robots!’ Amazon was to me one of the few places where there was just this very meaningful connection between robots and what the business value for the company was, and how we can really improve our associates’ experience."
"One thing I also really believe in is smart people will come with answers to questions, but it's really the questions that matter. And one of the nice things about being at Amazon is that I get to understand what the questions are and I get to frame the direct questions that I can then sort of unleash upon amazing brilliant people like you (Nia Jetter), to answer.”
Jetter: “The opportunity to help people, to obsess over our customers’ needs, to meet our customers’ needs, and solve some very real challenges that actually need to be solved in order to continue to meet our customers’ needs. Finding a way to truly help people here is something that is a huge attraction to me.”
On asking the right questions
Srinivasa: “There’s a lot that goes into building a product that is not science. A lot of startup founders or even technologists that I work with that say, ‘I’ve got this cool tool or this cool idea and we should do it.’ And it’s really about the what, why, when, where, and how. You could build a flying car and nobody might want it, history is strewn with examples of things that nobody wants, even though it was technically very hard to create.”
On being a science leader
Srinivasa: “Put yourself in a situation where your failure has material consequences. Whether you are a professor or whether you are a product leader, otherwise you are just dabbling. I’ve always rejected dabbling because I’ve always wanted to be in situations where the work that I did had real consequences, whether it succeeded or if it failed. That somehow really sharpens me and gets me excited about doing it. While it’s nice to have some safety nets, I do also think we should take the leap of faith and do something whose success or failure has material consequences.”
On loving robotics
Srinivasa: "The journey of being able to do it all. I love writing code, I love building robots, I love welding metal, I love proving theorems, but the opportunity to do it all and to really align against metrics and do it in a way such that you are able to bring meaningful change has always been really exciting for me."
On where they get their ideas
Srinivasa: “I love two things, one is observing the world and the other one is trying to explain things. I love explaining things to my kids, I love teaching. I think the act of teaching and the act of explaining really forces you to ask the five whys. I am very curious, I love reading about various things. Robotics is one of those things where you watch the world behave and you try to ask, okay, why is it behaving like the way its behaving and how do I think about it clearly? In many ways it is a very descriptive science. In that, when I look at a robot picking up a coffee mug, and I prove a theorem and I write an algorithm and build a robot that picks up the coffee mug, in some ways I'm explaining using the language that is available to me how a coffee mug is picked up.
“One of the projects I work on as a professor at the University of Washington is on a robot that can help feed people with disabilities. The reason I started working on that problem was because I visited the Rehab Institute of Chicago, it’s now called the Ability Lab, and I just asked people ‘What can I do that can at least attempt to make your life better?’ The top request from them was they just want to be able to eat by themselves and not have to be fed by a caregiver. So I was like ‘OK, I’ll do that, that sounds meaningful and important and I’m sure it’s challenging.’ History has shown that we’ve invented many things that we think are useful, but are not. So talking to customers is really, really valuable.”
Jetter: “I think of myself as someone who doesn’t just think outside the box, but exists outside of the box. I try to be very observant and I try to listen a lot and I try to draw analogies between experiences. I try to leverage some of my experiences that might be unique from my perspective, particularly coming from aerospace and defense and now being in robotics, just leveraging my past experience and bringing that new diversity of thought, in many ways, to robotics. I’m super passionate about fundamentals and first principles and breaking things down.
“A lot of my ideas come from drawing analogies based on my experience, so seeing something new that I might not be expert in or have depth in and relating it to something that I do have depth in and looking at it through a different lens. That’s been effective for me in at least a couple of instances in my career.”