AWS Announces the Deepracer League (DRL)
AWS Launches the First Global Autonomous Racing League, Open to Everyone
Today at AWS re:Invent, Amazon Web Services, Inc. (AWS), an Amazon.com company , launched the AWS DeepRacer League (DRL), the first global, autonomous racing league, open to everyone. Starting in early 2019, 20 AWS Summits worldwide will host a DeepRacer series tournament where anyone can compete in a race with their 1/18th scale AWS DeepRacer car driven by a reinforcement learning model built in Amazon SageMaker. Contestants can compete in as many events around the world as they wish, and the winners of each stage, plus the top 10 points scorers across the races, will qualify for the DeepRacer League Cup, held at re:Invent 2019 in Las Vegas, Nevada. Racers can also compete in virtual events and tournaments throughout the year by entering time trials on special tracks in the AWS DeepRacer simulator, available in the AWS Management Console. As with the physical events, winners and top points scorers in the virtual race circuit will advance to the DeepRacer League Cup at re:Invent 2019.
“By removing common challenges associated with reinforcement learning, giving developers the chance to have some fun, and providing them a complete autonomous model racing car along with AWS machine learning services like Amazon SageMaker, we are putting every developer in position to experiment with reinforcement learning and machine learning.”
The inaugural DRL event took place at this year’s AWS re:Invent conference in accelerated form over a duration of 22 hours. Starting on Wednesday afternoon, thousands of developers seized the chance to learn about reinforcement learning powered by Amazon SageMaker in workshops where they were also the first customers to receive AWS DeepRacer cars. At a specially-constructed racing area in the MGM Grand Garden Arena dubbed the “AWS DeepRacer MGM Speedway,” developers tested their reinforcement learning models and had their lap times with their cars recorded onto a Speedway Leaderboard. The fastest times advanced to the final where Rick Fish, co-founder of Jigsaw XYZ, from London, England, emerged victorious, taking the DRL Cup with a winning lap time of 51:50.
“Until now, developers interested in experimenting with reinforcement learning had to study academic papers and cobble together models with limited help. AWS DeepRacer and the DeepRacer League gives them the opportunity to discover reinforcement learning in a hands-on fashion and then proceed to build, train, and tune reinforcement learning models and deploy them into their autonomous model racing cars,” said Swami Sivasubramanian, Vice President, Amazon Machine Learning. “By removing common challenges associated with reinforcement learning, giving developers the chance to have some fun, and providing them a complete autonomous model racing car along with AWS machine learning services like Amazon SageMaker, we are putting every developer in position to experiment with reinforcement learning and machine learning.”
Reinforcement learning is a powerful type of deep learning capable of optimizing decisions in complex environments without requiring any labeled training data in order to achieve a long term goal. With reinforcement learning’s steep learning curve and blockers for adoption, AWS’s introduction of the AWS DeepRacer and the DeepRacer League is one more step in AWS’s mission of putting machine learning and reinforcement learning into the hands of everyday developers.
“When I first heard the announcement about AWS DeepRacer in the keynote, I was totally pumped and thought it was a great way to get people interested and started in reinforcement learning. It’s a field with an incredibly hard barrier of entry, and it poses a mental blocker, but AWS DeepRacer and DRL really opens it up for people,” said the winner of the first DRL Cup, Rick Fish, co-founder of Jigsaw XYZ. “Amazon SageMaker RL, the pre-built models, and available frameworks made everything really accessible, such that in less than a day I was able to have this fantastic outcome. As someone who has never worked with reinforcement learning before, I wasn’t expecting to qualify for the finals at all I thought someone was joking with me when I got the call! The whole experience was good fun, and I haven’t even begun to scratch the surface in terms of the service’s capabilities. I’m excited to personalize my car further, and I look forward to exploring SageMaker RL and AWS DeepRacer more.”