Jai Krishna BITS Pilani (Hyderabad, ‘18) graduate, with an M.Sc. in Physics and a B.E. in Civil Engineering, who has always been driven by a passion for robotics. His interdisciplinary expertise spans electronics, mechanics, and programming, making him a versatile researcher. Jai’s journey from app development in his school days to robotics research highlights his love for hands-on projects and real-world applications. Currently, he is focused on learning-based solutions for the loco-manipulation of legged robots in challenging environments. His work, particularly in the field of robotics, is fueled by a deep enthusiasm for deploying field robots that can navigate and operate in unpredictable terrains.

How did your involvement in the Automation and Robotics Club (ARC) shape your passion for robotics, and can you share some key projects and collaborations from your time at BITS Pilani?

My experience with ARC was instrumental in shaping my passion for robotics. When I joined, there wasn’t a robotics minor, and very few faculty members focused on the field. I often sought advice from Professors like Prof. Y. V. D. Rao and Prof. Arshad Javed for ARC-related projects. As the Joint Secretary, I helped start the autonomous underwater robotics team, and we initiated the underwater rover project through the Student Innovation Project (SIP), securing funding and faculty support. During COVID-19, we collaborated with Phoenix Professors and Yashoda Hospital to develop an oxygen flow regulator, which led to a published paper in IEEE.

In terms of projects, I worked on fun, self-driven initiatives like building a gaming wheel during winter break, which made racing games more engaging. I also created an interactive display and a light bar that reacted to phone notifications and music. These projects taught me CAD, electronics, and programming. Though I didn’t have many internships, I gained valuable experience through competitions like Flipkart Grid, where we built an autonomous stair-climbing robot. My hands-on work through ARC deeply enriched my learning and practical skills.

Were you always interested in robotics, or how did you develop the interest?

I started with app development in School, creating an app for the Microsoft Store around 7th or 8^th grade, maybe in 2015 or 2013. This exposure got me into computers early on. I was not very focused on the JEE preparation in School, and my Schooling was quite relaxed. A turning point was

participating in the ImagineCup, a Microsoft event for engineering students to showcase apps. I even met the Microsoft India CEO, got featured in the newspaper, and received a Lumia phone as a prize. That experience pushed me towards tech, and my interest gradually shifted towards robotics. The freedom my parents gave me allowed me to explore this field without any pressure. While I started with programming languages for app development, I eventually got bored and explored other areas like ethical hacking before landing on robotics. A local learning center in my city introduced me to Lego Mindstorm NXT kits, and I loved seeing my code interact with physical objects. This fascination, combined with support from seniors at BITS, solidified my passion for robotics, and I began exploring more advanced projects and learning through hands-on experiences.

Did having a low CGPA hinder you in your career prospects or during college life?

Absolutely, yes. I graduated with a CGPA of 6.34, with my lowest being 4.7 in one semester, possibly in my second year. During that time, I debated whether to apply for a technical secretary position, as falling below a 4.5 would have disqualified me from the role and even affected my ability to graduate. Ultimately, I did pursue the position, despite the challenges.

The key takeaway here is that a low CGPA doesn’t mean you’re out of options— - it just means you need to work harder than those with higher GPAs. For instance, my background is in Civil Engineering and Physics, but my interest lies in robotics. While my CGPA was low due to focusing on robotics, I showcased my projects and research work instead. This helped me secure a fully funded thesis in robotics and later, a job in the field in the U.S.

Though a higher CGPA would have made things easier, emphasizing other aspects of your profile can still open doors. If you can improve your CGPA, it’s worth it. But even if not, there are other ways to prove your capabilities.

Your thesis in Japan involved working on Quadruped Robots. Can you talk about what led to that?

My journey with quadruped robots started back when I was working with Abhishek Sarkar sir Sir on robotic actuators, specifically for a robotic dog project. My friends and I pitched the idea as a research topic. But we couldn’t afford to buy a robotic dog, so we decided to build one from scratch. During COVID, we worked informally, designing actuators using brushless DC motors with planetary gear systems and magnetic encoders. We built a single leg that could jump, but I had to pause when my thesis plans started coming together.

While I initially applied to a Japanese lab for humanoid robot research, they had just received a new robotic dog and asked if I’d work on that instead. It felt like a perfect continuation of my previous work, so I accepted. My thesis focused on applying deep reinforcement learning for quadrupeds in civil Civil Engineering tasks like inspection and surveying. Now, I work at Field AI, a robotics company deploying robots for safety and inspection in construction sites, and oil, and gas fields exactly what I envisioned in my thesis. It’s been a full-circle experience that I never anticipated.

Currently, you’re a visiting Deep Learning and Robotics Researcher at Field AI. Can you tell us about what your role involves?

I started as a classical robotics guy, working on underwater rovers and control systems. However, during my thesis in Japan, I transitioned into integrating machine learning with robotics. My Professor allowed me to explore machine learning on a new robotic dog, even though I had no prior experience. That’s when I learned to use ML/AI for robotics, specifically for locomotion.

At Field AI, my work focuses on teaching robots to navigate through varied terrains like stairs, uneven ground, and obstacles. Traditional methods require precise models of the environment, which can be slow to adapt. In contrast, learning-based methods use simulations to train thousands of robots to optimize their movements without needing explicit terrain models. For example, we define a reward system for moving forward, and the robots learn by trial and error in simulation.

Now, I work on both locomotion and manipulation teaching legged robots and humanoids how to walk, avoid obstacles, and use grippers to interact with their environment. It’s particularly useful in unpredictable environments like construction sites, ensuring robots can adapt and move without falling. This role has allowed me to apply my thesis experience directly in real-world scenarios.

Where do you see the future of robotics heading, and how can students find opportunities in this evolving field, especially for research positions abroad?

The field of robotics is evolving rapidly, with significant advancements in recent years. Companies like Boston Dynamics, which started with quadrupeds like BigDog for military use, have since developed more advanced robots like Spot and Atlas. Today, newer companies can achieve much of what Boston Dynamics’ robots can do but in a shorter time frame, thanks to the maturation of the robotics industry and innovations in AI. The integration of AI, particularly with models like ChatGPT, allows robots to interpret commands and perform tasks based on language prompts. The focus now is on transitioning robots from research labs to everyday applications, such as robotic taxis, warehouse automation, and personal assistants.

For students looking to break into this field, particularly in research, cold mailing is an effective strategy. When seeking research labs, I recommend looking at recent conferences and reviewing papers from the last two to three years. Identifying active researchers in your area of interest ensures you’re reaching out to labs that are currently contributing to the field. This method helped me find labs when I was pursuing learning-based robotics, and it remains a great approach for finding relevant opportunities abroad.

What are some of your fondest memories from hostel life, particularly any mischievous incidents with friends?

One of the funniest memories I have is from first-year exams, especially during Engineering Graphics (EG). Before every EG exam, there was always a power cut the night before— - it happened both semesters! It became a weird, almost-expected event.

Apart from that, I was heavily involved in department work, especially during my third and fourth years when I was the Technician Secretary. I enjoyed organizing events and working with NSS and ARC. For ARC, we used to stay overnight in the Sandbox lab, way past our hostel in time, which back then was 2 AM. The guards probably thought we were PhD students because we stayed late working on our projects. We’d sometimes sneak over to the Tinkerer’s Lab in SAC to grab parts, and by morning, we’d be so exhausted we’d sleep on the beanbags in the Sandbox. The mechanical department students would find us there and complain. It was all harmless fun, but those late-night sessions created some of the best memories!