SHREVEPORT – Autonomous racing cars, machine learning and artificial intelligence algorithms.
No, it’s not the 26th sequel of Fast and Furious -- it’s an opportunity for high school students at LSUS.
LSUS is joining the Amazon Web Services DeepRacer circuit, where students can develop a racing car algorithm and test it in a virtual and real-world environment, pitting their autonomous cars against competitors.
“This project is focused on teaching students how to train and use artificial intelligence by training a race car,” said Dr. Roger Zhao, an assistant professor in computer science at LSUS and director of this project. “Students will use a virtual track to test their model, and then we’re going to bring all the kids to campus and have a real race car competition on a physical track.”
Any high school student aged 14-17 can register for the free program at https://www.ce.lsus.edu/courseDisplay.cfm?schID=781. Students will be trained online throughout the fall and winter with the program culminating in an in-person spring race at LSUS.
Students start by customizing one of the AWS DeepRacer training models.
The model is powered by reinforcement learning, which is an advanced machine learning technique that can make short-term decisions while optimizing for a long-term goal.
These virtual models can prioritize speed, staying in the middle of the track or traveling in straight lines. The autonomous race car drives itself around the track based on its training model. Students will use software like SageMaker and AWS RoboMaker.
The more the model is tested in a virtual simulator or on a physical track, the better that model responds to the mission – racing around the track in the shortest amount of time.
The car uses a camera to help it navigate around the track, but the car’s path, speed and other facets are driven by the customized training model and reward functions.
These autonomous models can race against competitors in a web-based simulator as part of the AWS DeepRacer Student League.
Students’ models compete against other students from around the country and the globe on a virtual track.
But LSUS will also have physical cars and tracks available for in-person races, and the school aims to host an in-person race this coming spring.
AWS DeepRacer is a fun way to engage with reinforcement learning, which has proven effective in solving a wide range of autonomous decision-making problems.
“This is more than just autonomous racing – what these students are learning has applications in areas like financial trading, fleet logistics and data center cooling,” Zhao said. “This type of knowledge is highly sought after by many industries.”
Reinforcement learning has traditionally had a steep learning curve because of its extensive technological scope and depth, but AWS DeepRacer has created a controlled environment that allows students to introduce themselves to these concepts via autonomous racing.
Students can learn about creating and tweaking algorithms to achieve faster times on the track.
The venture is a partnership the LSUS Cyber Collaboratory and the Computer Science department.
No previous knowledge of the subject matter is needed to register and participate.
To watch AWS DeepRacer in action and learn more about the process, visit https://www.youtube.com/watch?v=JGMNk_HfruY.
