Abstract
Driving in a dynamic, multi-agent, and complex urban environment is a difficult task requiring a complex decision-making policy. The learning of such a policy requires a state representation that can encode the entire environment. Mid-level representations that encode a vehicle's environment as images have become a popular choice. Still, they are quite high-dimensional, limiting their use in data-hungry approaches such as reinforcement learning. In this article, we propose to learn a low-dimensional and rich latent representation of the environment by leveraging the knowledge of relevant semantic factors. To do this, we train an encoder-decoder deep neural network to predict multiple application-relevant factors such as the trajectories of other agents and the ego car. Furthermore, we propose a hazard signal based on other vehicles' future trajectories and the planned route which is used in conjunction with the learned latent representation as input to a down-stream policy. We demonstrate that using the multi-head encoder-decoder neural network results in a more informative representation than a standard single-head model. In particular, the proposed representation learning and the hazard signal help reinforcement learning to learn faster, with increased performance and less data than baseline methods.
Original language | English |
---|---|
Pages (from-to) | 701-710 |
Number of pages | 10 |
Journal | IEEE Transactions on Intelligent Vehicles |
Volume | 7 |
Issue number | 3 |
Early online date | Feb 2022 |
DOIs | |
Publication status | Published - Sep 2022 |
MoE publication type | A1 Journal article-refereed |
Keywords
- Automobiles
- Autonomous vehicles
- Hazards
- Image reconstruction
- Multi-task learning
- Policy learning
- Reinforcement learning
- Representation learning
- Task analysis
- Trajectory
- Vehicle dynamics