Matthew Tang of Irvine, CA, won the 2018 $2,000 Orange County Mensa Scholarship.
Our car had been following the orange line precisely, curving back and forth when suddenly it swerved off road! “STOP!” I slammed the kill-switch, and the robot car jerked to a halt. Locking in on anything orange, the programmed car had detected a citrus-orange shoe nearby and veered towards that instead. Luckily, no innocent bystanders were harmed in this trial run.
Ever since I spotted Google’s self-driving cars zipping around town last year in Mountain View, I marveled over the independence these vehicles can offer the elderly or disabled. I never dreamed that I could work on one as a high school student until I found the MIT Beaver Works Summer Institute. In Summer 2017, I was selected to collaborate with students from across the country on building robot cars about 1/10th the size of actual cars. During morning lectures, my intellectual curiosity was fueled digesting material adapted from collegiate robotics courses; in the afternoon, my computational thinking was challenged as we transformed theoretical concepts into functional code for the Rapid Autonomous Complex-Environment Competing Ackermann-Steering Robot (RACECAR).
Though I had prior programming experience with Python, Arduino and Matlab, designing the RACECAR in Robot Operation System (ROS) also required a solid grasp of circuits and systems to develop a control system by which to steer the car. The first step was to implement a Proportional, Integral, Derivative (PID) control system. I realized the calculus-based equation neatly simplifies to the following: Proportional is dependent upon the present error, Integral is dependent upon the accumulation of past error, and Derivative is the prediction of future error. The sum of all these terms is used as feedback to reduce error as quickly as possible without overshooting.
My team and I soon learned that properly tuning the constants was crucial to having a stable controller; in our first attempt, our car oscillated wildly since our implementation overcompensated for errors. We reevaluated and experimented until we found the optimal tuning constants which produced a stabling effect. When the car maintained a straight line, I proceeded to create a potential field controller to direct the car into open space and around obstacles since objects near the car exert a repulsive force inversely proportional to the distance from the car.
At the end of the four weeks, my team and 12 others witnessed the fruits of our labor put to test in a Grand Prix. I could hardly breathe as I watched the car make the hairpin turn, navigate over and under a bridge, and stay on a “yellow-brick road” surrounded by “blue water.” When the car accurately detected each change in the dynamic environment and switched algorithms seamlessly, I finally released a breath and cheered.
I have always had a passion for technology and engineering, whether it was combining nanotechnology with software and electronics to develop a long-term wireless heart monitoring system with novel dry electrodes or designing a 3D printed robot arm capable of grasping and flipping stacks of pennies. This summer program, along with conducting research in biomedical engineering at UCI Khine Lab and years of building and competing for Science Olympiad, solidified my desire to pursue computer engineering, specifically artificial intelligence. The study of autonomous intelligent systems will enable me to push the boundaries to solve some of the greatest challenges we face today.