In 2018, I sat in the audience at AWS re:Invent where Andy Jassy announced AWS DeepRacer, a fully autonomous 1/18 scale race car powered by reinforcement learning. At the time, I knew little about AI and machine learning (ML). As an engineer migrating from legacy networks to cloud technology, I never thought of myself as a developer. However, AWS DeepRacer immediately caught my interest with its promise of allowing even inexperienced developers to get involved in AI and ML.
The AWS DeepRacer League was also announced. It features physical races at AWS Summits to be held around the world in 2019, as well as virtual leagues in simulated environments. The winner will be eligible to compete for the title of Grand Champion in Las Vegas the following year. In 2018, AWS DeepRacer was just announced, allowing re:Invent attendees to compete directly at MGM Grand using pre-trained models.
My JigsawXYZ colleagues and I headed straight to the MGM Grand after the keynote. Despite the long line, we persevered while waiting and observing others racing. Participants selected a pre-trained model by answering questions about their driving preferences. Unlike later competitions, racers had to physically track down their cars and get them back on track when they veered.
We found that the model provided by AWS was unstable and slow by today’s standards, and frequently went off track. We concluded that we could get good lap times by quickly changing cars on the track. This strategy secured second place on the leaderboard.
The night before the finals, we learned that we had qualified due to elimination. Panic set in when I realized I would be competing on stage in front of thousands of people with little knowledge of ML. We desperately tried to train the model overnight to avoid embarrassment.
The next morning we were in the front row of the auditorium, next to Andy Jassy. Our boss, Rick Fish, represented our team. After a spirited introduction from IndyCar commentator Ryan Mylane, Rick posted a lap time of 51.50 seconds, earning him the title of 2018 AWS DeepRacer Grand Champion.
Image 2 – Rick Fish receiving the AWS DeepRacer trophy from Matt Wood
2019: Build your community and dig deeper
When I returned to London, interest in AWS DeepRacer exploded. We have spoken at multiple events, including hosting our own events. An evening with DeepRacer To gather. As the 2019 season approached, I needed to earn myself a spot in the finals. I started training the model in the AWS DeepRacer console and started experimenting with physical cars, including remote control and first-person projects.
At the 2019 London AWS Summit, I won the AWS DeepRacer Championship with a lap time of 8.9 seconds. This was a significant improvement over the previous year. This event also sparked the creation of the AWS DeepRacer community, which has since grown to over 45,000 members.
I became interested in understanding the inner workings of AWS DeepRacer. I contributed to open source projects that allow you to run training stacks locally and delved into AWS services such as Amazon SageMaker and AWS RoboMaker. For these efforts, I was named an AWS Community Builder.
I improved my skills in Python, Jupyter, numpy, pandas, and ROS by working on community projects. These experiences proved invaluable when I joined Unitary, an AI startup focused on reducing harmful online content. In less than a year, we built a world-class inference platform that processes more than 2 billion video frames every day using dynamically scaled Amazon Elastic Kubernetes Service (Amazon EKS) clusters.
Image 3 – Unitary demonstrating dynamically scaled inference with 1000+ EKS nodes at AWS London Summit
2020-2023: Virtual racing and continued growth
Due to the coronavirus disease (COVID-19) pandemic, the 2020 and 2021 AWS DeepRacer competitions have been moved online. Despite this, exciting events like the AWS DeepRacer F1 Pro-Am kept the community engaged. The hardware has been significantly upgraded with the introduction of AWS DeepRacer Evo with stereo cameras and LIDAR detectors.
In-person racing resumed in 2022 and I set a new world record at the London Summit. Even though I didn’t win the finals that year, the experience of competing and connecting with my fellow racers was still invaluable.
Images 4 and 5 – Author holding the 2022 London Summit trophy (left) DeepRacer community members and pit crew hosting the AWS DeepRacer workshop at re:Invent 2023 (right)
In 2023, competition is even more intense. In London, he set another world record, but it wasn’t enough for him to take first place. In the end, they effectively won the Europa League round and qualified for the final. Although our performance in the finals did not improve on our previous results, the opportunity to reconnect with the AWS DeepRacer community was worthwhile.
Conclusion: The lasting impact of AWS DeepRacer
Over the past six years, AWS DeepRacer has had a huge impact on my professional and personal life. This helped me build a strong foundation in AI and ML, improve my coding skills, and network with friends and professionals in the tech industry. The experience I gained through AWS DeepRacer directly contributed to my success at Unity, gaining recognition as a top UK startup.
As the official AWS DeepRacer League draws to a close, we can’t wait to see what the community accomplishes next. This journey shaped my career and life in ways I never expected when I first saw a small self-driving car on stage in 2018.
If you are interested in starting your own AI and ML journey, we encourage you to explore the AWS DeepRacer resources available on the AWS website. You can also join Discord’s active community to connect with other enthusiasts and learn from their experiences.
About the author
matt camp is an AI and ML enthusiast and has been involved with AWS DeepRacer from the beginning. He currently works at Unitary, applying his skills to developing cutting-edge content moderation technology. Matt is an AWS Community Builder and continues to contribute to open source projects in the AWS DeepRacer community.
