Bringing different elements of engineering together to create artificial life with a limited purpose seems like science fiction, but it is a reality today !
Karthikeyan Yuvaraj, our next pathbreaker, Roboticist at Intrinsic, an Alphabet company (Google’s parent company), works on problems in advanced robotics and perception for diverse applications in different industry verticals.
Karthikeyan talks to Shyam Krishnamurthy from The Interview Portal about his work on exploratory robots that can replace humans in potentially hazardous situations and industrial robots that can autonomously navigate into a trailer to unload cargo.
For students, being a roboticist is fun, but it requires you to spend long hours in the lab or on the factory floor working with these machines and getting your hands dirty if necessary !
Karthik, what were your initial years like?
I was born and raised in Neyveli, a small town in Tamil Nadu. My parents were government employees. Neyveli is a small mining town where power production is the primary revenue. Other than that, it is a tranquil town with nothing much to do. So if you are a kid, you can very easily get bored. Luckily I had some good friends to play with. There is a strong sense of community in the town. And everyone knows everybody, being a smaller community.
Math was my favorite subject in school. I was the school topper in math during my 10th standard public exams. During my 12th standard public exam, I was the state topper in math, securing 100 %. Other than math, I was very much interested in programming. BASIC is the first programming language I learned. These interests put me on track towards pursuing a computer engineering degree.
Thanks to my Father’s strong interest in science fiction stories and movies, I also developed a strong liking for them. Movies like Starwars, Robocop, and stories by Issac Asimov developed my fascination for robots and their role in society.
Also, in retrospect, living in Neyveli was like being in a steampunk world minus the Victorian aspect, because Neyveli, being a mining town, has bucket-wheel excavators which are visible throughout the city. A bucket-wheel excavator is a piece of massive machinery used for continuous digging in open-pit mining operations. In theory, these are giant teleoperated robots. At night, they would appear like a giant spaceship from a distance. When I was a child, I thought these massive ground-digging robots worked tirelessly throughout the night.
I was also quite interested in charcoal sketching. I especially loved sketching character concept art with subjects like weird humanoid creatures, alien beings, cyborgs, etc.
What did you do for graduation/post-graduation?
My passion for Mathematics and Programming led me to pursue a bachelor’s in Computer Science and Engineering. Based on merit, I got a seat at the Amrita School of Engineering.
Honestly, I did not find developing code for some banking application, or managing massive databases interesting. But at the start of the third year, I was introduced to embedded systems and computer architecture courses. These courses discussed programming components like motors, sensors, and light diodes in theory. Programming intelligence into such hardware components intrigued me a lot. This eventually led to a senior design project, building and programming an autonomous four-legged robot.
My “actual” college education began when I tried to build my first robot.
Bringing this senior design project to life required me to educate myself about soldering, designing PCB layouts, real-time programming, controlling a servo through pulse width modulation, etc.
During my senior design project, I decided to further my robotics studies. To specialize and have a career in robotics, I needed to go for higher education. Thanks to the support and encouragement from my parents, I applied to some schools in the USA for a masters in robotics. My dream was to pursue an MS in Robotics from the University of Pennsylvania. But I got rejected. I got accepted into its neighboring school, Drexel University.
During graduate studies, our lab had the opportunity to compete in the DARPA (Defense Advanced Robotics Projects Agency) Robotics Challenge. As a result, I made some conference and journal publications during my project with DARPA. Also, my master’s thesis was based on a humanoid robot driving a utility vehicle, which was one of the challenges in the DARPA competition.
What were some of the influences that led you to such an offbeat, unconventional and cool career?
It started with a desire to build my first robot. That led to making a miniature autonomous four-legged robot for my senior design project (during my bachelor’s). During that project, I liked how multifaceted the effort was. It involved software engineering, PCB design, embedded systems, etc. To put it in simple words, you bring different elements of engineering together to create artificial life with a limited purpose.
Looking back now, pursuing robotics at Drexel University is a significant part of my life, as it helped me compete in the world’s biggest robotics competition.
During my graduate studies, I sought Dr. Paul Oh to work in his robotics lab. After some months of initial assessment, I started to work in his lab as a graduate research assistant. Dr. Paul Oh is a renowned name in the world of Humanoid robotics. Working in his lab ( Drones and Autonomous Systems Lab) helped me to learn practical aspects of robotics, perception, and processing data from color based/depth sensors, and integrating such sensors into robots. His role as my mentor and thesis advisor was critical to starting my robotics endeavor in the USA.
After graduation, I began to work in Honeywell R&D. During my time at Honeywell, I got to work under Matthew Wicks, a very prominent name in the world of industrial robotics and automation. His passion and vision for modern robotics and its application greatly inspired me. In addition, I learned a lot of valuable lessons from him regarding the commercial aspects of robotic systems.
So it started with a fantasy of building robots. Then that led to educating myself in the world of robotics. But the primary catalyst was the mentorship and support from the folks I mentioned above, which kept me on course to pursue a career in robotics.
How did you get your first break?
Working as a robotics researcher for the DRC (DARPA Robotics Challenge) was my first break. The competition’s main goal was to design and build robotics systems that could neutralize nuclear disasters instead of sending humans. The inspiration for the competition was the Fukushima nuclear disaster, where many lives were lost during the effort to fix the situation in the nuclear plant.
When I was a graduate researcher in DASL (Drones and autonomous systems Lab), my professor got a grant from DARPA to compete in the DARPA Robotics Challenge. For this, he assembled a multi-disciplinary team of roboticists and computer vision researchers from several universities. Coincidentally, my effort in the lab during that time was based on visual servoing (vision-based robot control); hence I got the opportunity to work with the team. Or, in other words, I was in the right place at the right time :).
Competing in the DRC was once in a lifetime opportunity. We were competing with big boys like NASA, Boston Dynamics (which was acquired by Google then), The Robotics Institute from CMU, Lockheed Martin, etc.
Here is a photo of myself and the team members (covered by National Geographic) during the competition: https://newsblog.drexel.edu/2014/01/02/the-measure-of-a-robot-drc-hubo-put-to-the-test-at-darpa-robotics-challenge/#jp-carousel-2513
Also, here is a video regarding the competition: https://www.youtube.com/watch?v=oJakJQK21C0
After graduation, I completed two internships while still searching for a proper full-time position. One of the internships was at Children’s National Medical Center (in Washington, DC), where I was part of a research study to develop robotic systems for stitching during surgical procedures. Another internship was at the University of Pennsylvania GRASP lab, where the goal was to enable robotics to rehabilitate stroke survivors.
During the internship, Honeywell R&D (logistics and supply chain) was in search of a roboticist for their autonomous truck unloader team. So they reached out to me on LinkedIn. After the initial conversation over the phone, I was very much interested in their effort. Then, I further interviewed for the role in Saint Louis, Missouri. During the interview, my previous hands-on experience with robots, integrating camera sensors with robotic systems, and software development to enable intelligence on robots helped me get an offer.
The immigration rules might have changed now, but around 2013, after graduation from a university in the USA, one year of OPT (optional practical training) visa is provided, which gives the authorization to work. In addition, after one year, one can get a STEM extension of 18 months. I started to work with Honeywell on my OPT visa.
Around 2012-2013ish, it was possible to get a job in the robotics field. But the roles were based on traditional robotics built with PLCs (Programmable Logic Controller). Simply put, such positions focus on developing blind industrial robots programmed to repeat their hardcoded motions for a specific task. During my time at Drexel University, many seniors advised me to learn PLC and look for such roles, as they were easy to find. That was good advice (given the time), but such technology didn’t interest me.
And even today, PLCs have their place in modern robotics due to their stability/safety. And other technologies like AI, ML, IoT, etc., won’t make PLCs obsolete yet. But the role of PLCs in modern robotics is a highly talked about topic in the industry, which requires in-depth discussion, and we won’t be getting into that here.
Again around 2012-2013ish, for modern robotics, there were a lot of research engineer-based and post-doc positions in academia, but only a handful in the industry. Moreover, even in the industry, the applications were primarily military, and internationals don’t qualify for those roles. But during that time, some startups in silicon valley, Boston, and some MNCs were starting to explore such modern robotic solutions and needed roboticists. And I got an opportunity with Honeywell at that time.
But today, it is a very different picture; there are plenty of opportunities worldwide for various robotics sectors. Silicon Valley, Boston, and Texas have plenty of job opportunities in the autonomous driving and modern industrial robotics space in the USA.
Tell us about your career path in the field of Industrial Robotics
Working on the project with DARPA was a turning point in my career, as it led to many exciting opportunities. On a higher level, the technical problem to be solved for DRC was to design and build a robotic system to make intelligent decisions based on the data from its vision sensors, force-torque sensors (on its hands), and other onboard sensors. For example, if a robot needs to climb a ladder, it first needs to locate the ladder (among different things in view), head to the ladder, then start climbing (by planning the suitable motion). Or, if a robot needs to cut a hole in the wall, it first needs to locate where the drill is, grasp it in the right way, and correctly manipulate the drill to cut open a hole. This aspect of robotics was very intriguing, where the vision sensor data (combined with other sensory inputs) is processed to enable a robot to manipulate its environment.
Around this time, there were a lot of similar efforts in the automation space as well, where the goal was to enable blind industrial robots to see and make intelligent decisions. My experiences working in DASL (Drones and Autonomous Systems Lab) and DRC led to a robotics software engineer role in Honeywell R&D. This was my first foray into commercial robotics after transitioning from the world of academics/research-based robotics.
In Honeywell R&D, the main project I worked on was building an autonomous truck unloader. These are massive tank-like robots with dual arms. The goal is to make this robot autonomously navigate inside a trailer, locate the cargo, and start unloading it. This was an immensely multi-disciplinary project comprising engineering experts from different backgrounds. My role in this effort was to design a vision system to help with autonomous navigation and also localize the cargo inside a trailer. Working on the autonomous truck unloader was a worthwhile project, as it taught me the engineering requirements and skills to build a robotic system that can perform 24/7 on a customer site. Also, when working on this effort, I got several patents based on the vision system of this robot.
Not just Honeywell, many companies worldwide have started to focus on advanced robotics for the next industrial revolution, aka Industry 4.0.
One of the main buzzwords in Industry 4.0 is AI (Artificial Intelligence).
Robotics and AI are like the two sides of a coin, and I was very interested in working on more AI-involved robotic systems. In silicon valley (California), an AI startup called Vicarious was hiring roboticists to develop advanced robotic systems for industrial applications. I was one of the initial roboticists they hired.
Working in a small robotics startup in silicon valley was pretty challenging.
Though there is a lot of promise in AI-enabled robotics systems, it is quite a lot of work to get such systems working consistently and in a scalable manner. Vicarious had world-class computer vision researchers and roboticists who are experts in domains like motion planning, perception, neural networks, robotic object manipulation, etc. Working on such advanced technology with world-class experts in Silicon Valley is a dream come true for me.
The logistics and supply chain industry had a lot of interest in Vicarious’s advanced robotics/AI systems. That led to many customer deployments.
One of the main lessons I learned in Vicarious is what it takes to build an AI-enabled robotic system to work 24/7 in the real world alongside human labor.
Alphabet acquired Vicarious in April 2022. More on it here: https://techcrunch.com/2022/04/22/alphabet-owned-intrinsic-is-acquiring-fellow-robotic-software-firm-vicarious/
What were some of the challenges you faced? How did you address them?
Challenge 1: The naysayers
In India, the universal rule for a computer science graduate is to get an IT job. Even graduates from disciplines like Chemical, Mechanical, and Electronics/Instrumentation take the option to be placed in an IT company. Hence a computer science graduate looking for a role in robotics was a fish out of water scenario, and I could not find any such prospects in India. This may not be the reality now, but it was so in 2010 when I graduated. I was the only one in my computer science undergraduate batch who wanted to pursue the robotics track, while others stuck to an IT career. And it was very daunting and filled with a ton of uncertainty. There were questions and negative comments like, “why waste four years of undergraduate education if you wanted to get it into robotics?”, “Only US citizens or green card holders can get a job in robotics; international folks cannot make it into robotics unless they are pursuing PhD.” While pursuing my master’s, my seniors constantly reminded me that there is not much future in robotics for internationals. Seniors who came to Drexel to pursue the robotics track changed tracks to computer science or other fields that had a straightforward job market. There was one desire which was helping me to cope with all this negativity, and that was just the desire to build and spend time with robots. Deep down, I knew robotics/AI is the next significant invention humanity would need (similar to Electricity). I needed to get on this train, even though it wouldn’t be a pleasant ride. Luckily, around 2012ish, many advanced robotics-related efforts started to happen throughout the USA, and roboticists were in high demand (still are 🙂 )
Challenge 2: Not a walk in the park
Apart from the meta challenges to becoming a roboticist, working in the robotics/AI field itself is pretty challenging. Robotics is the most rewarding as well as the most frustrating field to work in. Most of the robotics/AI efforts are part of the innovation/R&D team in companies, and these are not typical 9 to 5 jobs. Being a roboticist, you are not expected to sit in your fancy cubicle, expecting things to happen; it requires you to spend long hours in the lab working with these machines and getting your hands dirty if necessary. Regarding robotics, one needs to cultivate a “boots on the ground” mindset and not be reluctant to spend hours on the factory floor or in the lab to be productive.
Challenge 3: Technical communication
Also, working in robotics requires strong technical communication because you will work with professionals from different backgrounds to make things happen. Communicating your problem or solution to someone from a diverse professional background is essential here. Since English was not my first language, it was hard for me to communicate effectively with my peers. This kind of problem can be solved by the trial and error method and by constantly asking for feedback. Yes, some will humiliate you in this process, but most will give constructive feedback, which helps. The worst thing you can do in such scenarios is to stop communicating and be in the shadows.
Where do you work now? What does your current role entail?
As of today, I work as a Roboticist at Intrinsic, an Alphabet company (Alphabet is Google’s parent company). The goal of Intrinsic is to democratize the potential of industrial robotics and make it accessible for millions of businesses, entrepreneurs, and developers worldwide. My role in the company is solving problems in advanced robotics and perception for applications in different industry verticals.
Being a roboticist in perception, one needs to have sufficient knowledge of robotics, computer vision, 3D geometry, and software engineering. The best way to learn the necessary skills for robotics is by getting involved in hands-on projects, just reading theory in isolation is a waste of time.
The challenging and fun part of being a roboticist is working with the robots and programming them to impart them with intelligent behavior. And most rewarding is when I see my robots working on the factory floor, helping folks with their day-to-day tasks.
How does your work benefit society?
“A robot goes where no human has gone”:
When I worked on the DRC (DARPA Robotics Challenge), the goal was to demonstrate how we can use robots to neutralize nuclear disasters instead of sending in humans, which can be very hazardous and even fatal. This kind of explorative robotics technology can be used for verticals like deep sea exploration, fire fighting, planetary exploration (like the mars rover), military, volcanology, archaeology, etc.
Eliminating mind-numbing repetitive work:
“Robots will take our jobs” is a ubiquitous quote. But, in reality, robots will help humans to have better jobs and work-life balance. Most of the labor force working in factories and delivery stations is being made to do repetitive jobs. Being up on your feet for 10-12 hours at a time and doing repetitive motions like sorting, lifting heavy boxes, etc., takes a toll on a person’s body over time. The robotics systems I have developed in the logistics and supply chain space can help take over these mind-numbing and repetitive jobs.
Imagine ten people working in an assembly line, packing 100s of boxes daily. Eventually, this can transform into nine robots working in an assembly line and one person managing it. Also, for this person to supervise these robots, it would require education to operate computers and robots (if necessary). Hence as you can see, this will help enhance the labor force to gain more skills than just being employed for physical work.
Tell us an example of a specific memorable work you did that is very close to you!
Working on the robotic truck unloader in Honeywell has been the best experience of my career so far. If interested, here is an article on it:
The robotic truck unloader is the biggest robot I have ever worked on. This massive tank-size robot autonomously navigating into a trailer and unloading cargo with its huge arms was genius, dangerous and weird. It felt like being part of a sci-fi movie.
Your advice to students based on your experience?
I am not here to say that robotics is the best and most fulfilling career one can have. However, building robots has been my calling since childhood; thanks to luck and hard work, it came to fruition. So my advice is, when you are young, try to figure out what’s your calling. All of us have different interests and tastes, but there will be something on which we want to build our life, and so you should religiously try to find that.
And during your pursuit, make sure you are surrounded by people who support you and also the people who discourage you. Because the more I was ridiculed or told that I would fail, the more I wanted to succeed.
And for future roboticists, the field is much more booming now than when I started. There are tons of resources in India and worldwide for one to become a roboticist. Also, for any guidance to get you started on this journey, please reach out to me at firstname.lastname@example.org.
For now, I want to continue pushing the envelope of developing and applying advanced robotics in different industrial verticals. In the next 3 to 4 years, I see myself further specializing in the perception aspects of robotics.
As mentioned before, being an avid sci-fi fan, I think I also have my own stories to tell. So, in the coming years, I will give a genuine shot to publish my very first sci-fi stories.