New AWS DeepRacer Student League will bring competitors to Ottawa for a wildcard championship race
Canadian students, 16 years of age and older, are competing in one of the world’s most interesting competitions this month. No, this isn’t a varsity sports championship. Instead, competitors are racing tiny, 1/18th-scale autonomous race cars in the new AWS DeepRacer Student League. Amazon Web Services (AWS) launched the global virtual league this March as a way of teaching students critical artificial intelligence and machine learning skills in a hands-on way.
“At AWS, we believe machine learning is one of the most transformational technologies of this generation,” said Mike Miller, General Manager of AI/ML Thought Leadership for AWS. “We want to ensure that anyone interested in technology has a fun and easy way to quickly learn these skills.”
What makes AWS DeepRacer clever is that competitors are racing cars and all the glory that comes with achieving faster lap times, but they are actually doing it by using a machine learning concept called reinforcement learning. Through online tutorials and a 3D virtual race simulator, students create a set of instructions called a ‘reward function” which tells the vehicle what to prioritize as it learns to drive around the virtual track, using trial and error to achieve faster lap times. This trial and error type of training forms the basis of the reinforcement learning model. For example, a car could be told to prioritize speed or stay close to the centre line of the racetrack. It also learns to avoid obstacles like other cars or debris.
For the league, each student receives 10 free hours of race car training each month to make their vehicles even better, because with reinforcement learning the cars themselves are collecting experience and data that help them automatically improve each time they drive around the track. This also allows racers to troubleshoot problems with their DeepRacer car while observing its movements and capabilities.
Once a student has trained a lean, mean, tiny racing machine, they’re ready to compete. Students can submit their models to the AWS DeepRacer Student League to race on new track designs and race challenges each month.
Miller describes the educational benefits of the program, saying “what’s exciting about the AWS DeepRacer program is that it allows students with no prior machine learning experience to go from first-time developer to training and racing their own reinforcement learning models very quickly. Competing in the league just accelerates the learning as students are driven to keep tinkering and trying new ideas in order to finish first. The skills students learn by racing these race cars for fun could help them solve some of the world’s biggest challenges one day.”
According to the World Economic Forum, 58 million jobs will be created through machine learning in the coming years, but only 300,000 engineers have the training to build and deploy learning models.
Previous AWS DeepRacer League events saw more than 70,000 developers take part from 150 different countries and 700 different organizations. While the AWS DeepRacer Student League is currently virtual, Canadian students will have a chance to watch their vehicles race in person at the AWS DeepRacer Canadian Wildcard Championship this May in Ottawa. Students will mingle with elected officials and policymakers while competing for prizes. In order to be eligible, competitors must sign up for the AWS DeepRacer Student League and submit a model before the end of March, when AWS will select the teams that will go to the nation’s capital.
“In the past decade, we’ve witnessed a huge shift in our economy to cloud and advanced technologies. Cutting-edge technologies, such as artificial intelligence and machine learning, are disrupting every industry from healthcare to automotive and within the public sector,” said Nicole Foster, AWS Director Global AI and Canadian Public Policy. “Today, digital skills are essential skills!”
Register to be part of the AWS DeepRacer program here: https://aws.amazon.com/deepracer/student/
