Su Pang earned his doctorate in electrical and computer engineering from Michigan State University. At MSU’s Connected and Autonomous Networked Vehicles for Active Safety, or CANVAS, Pang’s research focused on perception, or helping autonomous vehicles to see and sense their surroundings. He recently began a new role as a perception machine learning engineer at Nuro, a self-driving delivery robot company based in Mountain View, California.
I always liked cars as a kid and have always been interested in technologies related to cars, especially autonomous driving. When I was looking for Ph.D. opportunities, the MSU CANVAS program caught my eye. I really enjoyed the work I was able to do there on self-driving cars and am glad to see my work deployed on real test vehicles at MSU.
A lot of things attracted me to MSU, including the faculty who have strong academic backgrounds. I worked closely with Hayder Radha and Daniel Morris, who were wonderful advisers and helped me a lot both in my research and my life. The research platforms at MSU are wonderful. We have two self-driving test vehicles, which not many schools have, and we can collect data and do experiments using a real vehicle on the road, not just through simulations. We also have large test fields on campus, which gave me a lot of hands-on experience.
My research at MSU mainly focused on the perception module of self-driving systems, including 2D and 3D object detection using multiple sensors including camera, lidar and radar, multi-object tracking and sensor fusion. Thanks to the mobility-related research ecosystem at MSU, our team and I could build a real self-driving vehicle by ourselves and test the technologies that we developed on the road.
Perception helps a vehicle to see and sense its surroundings. This is a big and challenging topic for self-driving vehicles as many downstream applications are built on it, such as path planning and control. One cannot drive safely without clearly “seeing” the road and the surroundings. The self-driving industry also requires a lot of people to work on perception, so my doctoral research experience fits this very well. Through my program of study, I gained a good understanding of self-driving systems, their main challenges and how to help the community solve these problems.
My research experience helped me fit seamlessly into my new role as a perception machine learning engineer at Nuro. My job is to use machine learning to solve real-world perception problems, such as object detection, for our self-driving delivery vehicles. My work is very close to my research at MSU. The hands-on experience I gained working in the CANVAS program and the research I did with my advisers were all helpful for me to prepare for my new role. Compared to other candidates, I had an edge in terms of knowing much more about self-driving systems, what the challenges are and how to deal with them, and what is state-of-the-art.