What will we be doing
As a team, we will be working together on developing an algorithm that will make the DeepRacer car to be fully autonomous using machine learning, especially reinforcement learning. By autonomous, it is ment that the vehicle will use its sensors (camera) to follow a path- which is a racing track- without getting out of the boundaries and finishing the racing circuit faster than any other competitor. These circuits are physical as well as virtual, so if it is preferred to work out everything online we can! This means that the race would be a simulation using AWS DeepRacer Reinforcement Learning to train a virtual model. However, there may be competitions later that would require physical participation.
What is the purpose of this project
The purpose of this project is to automate the DeepRacer car so that it completes a racing circuit (Which will have any configuration) faster than any other competitor in the same circuit, always sticking to the rules. By doing this we will learn to work as a team and develop our understanding and knowledge on machine learning which is a very powerful aspect in autonomous decision making. This is a great opportunity for all participants to get started into reinforcement learning with one of the most flourishing companies on the globe.
Team's Description
Diego Padua:
Working with the AWS DeepRacer was a fun experience because I got the opportunity to learn a little bit about artificial intelligence and do hands-on work with it. What's more, I was surprised to learn that there are many different kinds of artificial intelligence, which as a part of the AWS DeepRacer we were mainly focusing on reinforcement learning. To clarify, reinforcement learning is a training method where you reward and punish the agent (AWS DeepRacer) based on how it interacts with its environment. As a part of this process, it was challenging because we had to experiment with a reward function where we decided on what algorithm to use or write that could help the car learn better in its simulated environment. Although, it was a trial-and-error process, I found it highly enjoyable, especially when I was competing with my club members to see who would break the record for the fastest track completion time. Aside from that it was also a very engaging process because we got to set up the track and learn how to calibrate the car using our model that was trained in the simulated environment. In essence, I found the AWS DeepRacer club project fun because it gave me a little bit of hands-on experience about artificial intelligence and even taught me a few things that I did not know before.
