End-to-end autonomous driving based on machine learning
In this project, we will explore the end-to-end approach to autonomous driving through machine learning. This project will be executed on the Donkey Simulator, a remote-control car simulator built on the Unity game platform. We will focus mainly on reinforcement learning algorithms as our primary ma...
Saved in:
| 主要作者: | |
|---|---|
| 其他作者: | |
| 格式: | Final Year Project |
| 語言: | English |
| 出版: |
Nanyang Technological University
2023
|
| 主題: | |
| 在線閱讀: | https://hdl.handle.net/10356/167648 |
| 標簽: |
添加標簽
沒有標簽, 成為第一個標記此記錄!
|
| 機構: | Nanyang Technological University |
| 語言: | English |
| 總結: | In this project, we will explore the end-to-end approach to autonomous driving through machine learning. This project will be executed on the Donkey Simulator, a remote-control car simulator built on the Unity game platform. We will focus mainly on reinforcement learning algorithms as our primary machine learning method. This report seeks to run simulations on autonomous driving based on selected on-policy and off-policy algorithms. These methods are namely, Proximal Policy Optimization (PPO) for the on-policy method and Soft Actor-Critic (SAC) as our off-policy method. In addition, we will be exploring the use of entropy regularization by passing different entropy coefficients while training our models to explore its effects on training. Rewards to the model are calculated based on the speed of the agent and the ability to stay in the centre of the track. The results from the trained models produced by these algorithms will then be drawn into comparison, based on the cumulative rewards of each algorithm. |
|---|