During my second year of university, I dedicated all my spare time to an entrepreneurship competition-CICSIC. The project I worked on was a distributed drive chassis motion controller for electric vehicles, led by Dr. Y from the Department of Vehicle Engineering (who was still a candidate at the time).
Although I frequently doubted my abilities and even considered dropping the project, I persevered. Through the process of market analysis and real-world vehicle testing, I developed a deep interest in autonomous driving and computer vision. The timeline for this project was from October 2021 to September 2022.
This project was undoubtedly fascinating #
My involvement in this project was made possible by my role as Vice Chairman of the Entrepreneurship Service Department at Southeast University, where my colleagues assisted Dr. Y in organizing an information session.
When I learned about the project, I had just spent a year in Nanjing, having left my hometown. Only about 5% of the people born in my hometown had received college-level education. As a result, when I entered university, I had little understanding of concepts like scientific research, markets, etc.. Dr. Y’s presentation opened my eyes to the tremendous power of scientific research. They were looking for a few undergraduate students to conduct market research, so I contacted them to express my interest in joining.
Treat the project as a GAN #
The most challenging phase was at the beginning of the project. I still adhered to my high school habits: waking up at 6:30 AM, doing course assignments, and previewing lessons—much like in fully-supervised learning, where the course syllabus served as my training label. I even experienced overfitting—relying too heavily on the knowledge provided by the course, without knowing how to acquire additional information.
During the first presentation, I was harshly criticized by Dr. Y and another master’s student. Rather than conducting market research, I was merely imagining a market. This failure led me to stay up all that night, learning knowledge on vehicles, control, and optimization.
This experience was more like training Generative Adversarial Network—I constantly deepened my understanding of the electric vehicle chassis controller market and presented more convincing business plans, while the ‘discriminators’—from PhD students to judges to investors—continued to raise their expectations. I have learned how to receive criticism and be humble about unfamiliar field.
Learned time management and developed a passion for AD and CV #
Southeast University and the School of Mechanical Engineering placed great emphasis on our project, as the competition involved not only inter-school rivalry but also inter-departmental competition. The pressure we faced was immense, especially for me to start from zero as a second-year undergraduate. Although I started off at a disadvantage, I have a personal trait: perhaps my initial efforts were subpar, but I always maintained a positive outlook on the outcome, which made me resilient in the face of setbacks.
I began to develop time management skills. While studying my major courses, I gradually increased the time I dedicated to the project from three hours per day. From April to August in 2022, I slept only five hours most of the days. After dinner, it was time to collaborate with teammates, whether it was conducting market research or performing real-vehicle testing. Often, we worked until 3 or 4 a.m., and then managed to squeeze in a few hours of sleep before catching up with our academic coursework. Fortunately, after years of supervised learning, my under-specification, course-oriented brain was still able to keep pace with my major courses.
My understanding of automobiles gradually deepened. I learned about distributed drive, domain controllers, hydraulic and electric drive systems, as well as perception, decision-making, and vehicle-to-everything coordination. I began to see these components as modular building blocks, and their engineering principles, which treat the design process with a sense of artistry, deeply impressed me. I also gained insight into the entire automotive ecosystem—OEMs, suppliers, and the broader industry landscape—which solidified my belief in the transformative potential of autonomous driving to revolutionize human life.
A precious memory and a valuable experience of growth #
Our project ultimately won the national silver prize, which was quite a remarkable result. More importantly, this project sparked my interest in autonomous driving, motivating me to explore machine learning and deep learning after the project’s completion, in early 2023. Furthermore, after spending half a year abroad in the United States, I officially began working on autonomous driving perception in February of 2024.
In addition to Dr. Y, Mr. D, Ms. Y, and Ms. S were also of great help to me. The year I spent with them was the most memorable year of my undergraduate studies, and the experience of collaborating with these four colleagues will remain with me for a lifetime. What stands out the most is the experience of watching the sunrise together after an all-night session at the mechanical building, revising our business plan.
Reflection on the Experience #
Sometimes, I ask myself whether I could have developed a deeper understanding of this field had I started working in a laboratory during my sophomore year. However, the truth is that my interest in autonomous driving gradually formed through this project. Before this, I had been short-sighted and had no understanding of graduate studies, autonomous driving, deep learning, or computer vision. I might have even gone straight into work after earning my bachelor’s degree, so the idea of joining a lab for research during my sophomore year was beyond my consideration.
#article #EnglishLast modified on 2024-12-03