APU-Amazon Web Services (AWS) DeepRacer Competition 2023 Showcases Future Machine Learning and Data Science Talents

Imagine racing an autonomous car on a physical track while learning more about Reinforcement Learning (RL), an advanced Machine Learning (ML) technique. This dream turned into reality for racing enthusiasts who raced on-campus on 13 April at Asia Pacific University of Technology & Innovation (APU), the first university in Malaysia to collaborate with Amazon Web Services, the world’s most comprehensive and widely adopted cloud offering, to host an AWS DeepRacer Competition.

The AWS DeepRacer is an autonomous 1/18th scale race car powered by AWS, designed to provide everyone with a fast and interesting way to learn about training reinforcement learning models, experiment with the technology by building autonomous driving applications, and test their RL models by racing on a physical track. The event attracted the participation of 215 teams comprising 535 future data scientists and engineers currently studying at APU.

The APU-AWS DeepRacer Competition 2023 was organized by the university’s Centre of Technology and Innovation (CTI), which is responsible for developing technology applications and solutions deployed across APU.

CTI has been collaborating with AWS to migrate APU’s application architecture to the cloud as part of the university’s ongoing Digital Transformation strategy, which aims to transform APU's IT infrastructure into a more agile, efficient, and reliable ecosystem that delivers exceptional learning experiences to students and faculty.

The month-long competition kicked off in March with AWS experts providing APU participants with a console walkthrough and a hands-on machine learning workshop. Contestants had the opportunity to test their skills by experimenting with multiple sensor inputs, the latest reinforcement learning algorithms, neural network configurations, and simulation to real domain transfer methods before testing them in autonomous car racing.

To further hone their technical knowledge and skills, the semi-finalists also attended a two-hour coaching clinic where AWS experts were invited as guest speakers to address technical questions, followed by the submission of their racing lap times.

The top 10 finalists were then identified to participate in an exhilarating two-hour Grand Finale. Teams “The Cloud Champs”, “Machine Meowers”, and “Common Emitter” emerged as the top three winners, with lap speeds of 8.885, 9.106, and 9.59 seconds respectively.

Mr. Julian Lau, Head of ASEAN Growth Markets, Worldwide Public Sector, AWS, joined participants at the event and spoke about the collaboration with APU. “Artificial intelligence (AI) and machine learning (ML) are among the most transformative technologies that can help us address society’s challenging problems. We are pleased to work with APU to provide its students with the opportunity to experiment with AWS’s ML services and develop ML skills in a fun and engaging way. Through the AWS DeepRacer League, we aim to inspire APU students to explore the full potential of ML and cloud technology, and apply them to real-world scenarios and their future careers.”

As part of the APU-AWS DeepRacer Competition 2023, each member of the finalist teams was given a complimentary 30 hours of simulation training with an AWS console, and 5 GB of model storage to train a reinforcement learning model in a simulated 3D racing environment.

The champions of APU AWS DeepRacer Competition 2023 attributed their success to their research and learning from others' experiences. Tang Jian Shiun, a Bachelor of Computer Science (Hons) Intelligent Systems student who paired up with his classmate Tey Jia Yi for the competition, said, “Throughout the training process, we emphasized the stability of the model car rather than focusing on speed. We are delighted that this approach awarded us first place.”

Chong Jack Pang, the team leader for “Machine Meowers” and his team members Tan An Ning and Wong Kai Ze, Intelligent Systems Year-2 undergraduates and second-place winners, credited their achievement to the knowledge-based theory they learned from the programme at APU. “We have learned machine learning and reinforcement learning in the classroom, and the familiarity undoubtedly helped us in this competition. Though this was our first competition, we made it to second place.”

“Common Emitter”, a team consisting of Bachelor of Computer Engineering with Honours Year-2 students Lee Zhan Tom, Wong Jun Zhi, and Yoong Ming Jun, emerged as the third-place winner in the highly competitive competition. Despite not specializing in machine learning and reinforcement learning, the team's proficiency in basic Python coding enabled them to excel in the machine learning and reinforcement learning process.

“We were able to perform well by utilising a reward-penalty mechanism to train the machine, which was run by an AI system. We need to train them in good habits to reach the objective set – to move as fast as possible without going off-track. If it makes the right move, we will reward it with high marks, but if it makes a mistake, we will cut down the marks as a penalty. It is pretty much like the carrot-and-stick approach,” explained the team.

Mr Gurpardeep Singh, Chief Operating Officer of APU, lauded the participants' active participation and learning enthusiasm, highlighting the value of the competition in providing real-world learning experiences. “APU-AWS DeepRacer Competition 2023 not only serves as a race but also equips students with practical knowledge and skills in working with virtual models using AI and reinforcement learning and transforming these models into real-life physical ones. Such experiences prepare students for their future professional careers, as they gain hands-on experience with cutting-edge technologies.”

The success of the APU-AWS DeepRacer Competition 2023 underscores APU's commitment to fostering a culture of innovation and providing opportunities for students to excel in emerging technologies. Congratulations to the winning teams on their remarkable achievement!