Hierarchical-Program-Triggered-RL
This folder contains the experiments and code for Hierarchical Program Triggered RL paper (https://ieeexplore.ieee.org/document/9497870). Entire framework is built on CARLA version 0.9.8 : https://carla.org/2020/03/09/releas and Python 3.6 and above. The sub dependencies are mentioned inside each folder.
Folder Structure
Training
Contains code for training the Deep Q Learning and the Deep Deterministic policy gradient agents.
Testing
The Testing of HPRL agents have been conducted on NHTSA pre crash scenarios modelled in Scenario Runner 0.9.8
Scenario Runner documentation : https://carla-scenariorunner.readthedocs.io/en/latest/ List of Supported Scenario : https://carla-scenariorunner.readthedocs.io/en/latest/list_of_scenarios/
| Control Loss | No Signal Junction |
|---|---|
 |
 |
| Vehicle Passing | Obstacle Avoidance |
 |
 |
Verification
Verification has been conducted by using Nagini : https://github.com/marcoeilers/nagini.
Videos
Contains the experiment simulation videos.