Three UCCS students took the top spots in the DeepRacer competition at the Space Symposium earlier this month.
The DeepRacer competition is hosted by Amazon Web Services (AWS) and Lockheed Martin to create more interest in machine learning. Students use machine learning to train a reinforcement learning agent – in this case, a car – to navigate a short closed course track. Training is done by simulation with the simulated car driving on a simulated track. For the competition, the trained model is transferred to a physical car to drive the physical track. Participants have the ability to modify the reward function and the simulation environment for training.
“The UCCS teams came from the Engineering and Applied Science Senior Design program, consisting of Data Analytics and Systems Engineering, Electrical and Computer Engineering, and Mechanical and Aerospace Engineering students who did great work supporting each other, resulting in 4 of the top 5 finishers being from UCCS,” said William Michael, an instructor in the College of Engineering & Applied Sciences. “Each of the curricula contributed in its own way, with coding being an important component in the exercise.”
Students were given ten hours of training time to test and allow the program to learn and coordinate around a specific track as efficiently as possible. The students used Python as the directive to command how the racecar would move.
“Essentially the program/agent initiates an action, like going through a race track (the environment), and tries to complete it,” explained Long Tran, one of the students who won the competition. “The first try will be difficult to complete, since the program does not know how to truly maneuver around the track aside from specific commands like steering and trying to stay between the lines. To teach the program, a reward function is implemented so it can teach the program to do the correct actions and not veer from the path. It takes some training for the agent to learn and eventually it can complete the task.”
Tran described the competition as a fun opportunity to learn something new.
“Machine learning is pretty advanced and being able to learn and practice in a controlled environment and physically see the process and outcome was a great experience,” he said. “It also provided the chance for participants to be able to go into the Space Symposium, which I had wanted to visit for years.”
Tran, who hopes to work in the aerospace industry after graduation, says the Mechanical and Aerospace Engineering program and especially the faculty of the department have been “a great resource to learn from and develop my critical thinking, and increase my capacity to learn new concepts.”
About the UCCS College of Engineering and Applied Science
The College of Engineering and Applied Science enrolls more than 1,700 students and offers 23 engineering and computer science degrees, ranging from bachelor to doctoral. The college is a Department of Homeland Security / National Security Agency Center of Academic Excellence in Cyber Defense and works closely with the National Cybersecurity Center and with more than 250 aerospace and defense, information technology, cybersecurity and engineering organizations in the Pikes Peak region. Learn more about the College of Engineering and Applied Science at UCCS.