# WorldOnRails

**Repository Path**: kin-zhang/WorldOnRails

## Basic Information

- **Project Name**: WorldOnRails
- **Description**: 自用.... 原链接请自行去github，此处为kin的探索repo（gitee快一点）
- **Primary Language**: Unknown
- **License**: MIT
- **Default Branch**: release
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2021-06-17
- **Last Updated**: 2022-05-25

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# World on Rails

![teaser](assets/teaser.jpg)
> [**Learning to drive from a world on rails**](https://dotchen.github.io/world_on_rails/)    
> Dian Chen, Vladlen Koltun, Philipp Krähenbühl,        
> _arXiv techical report ([arXiv 2105.00636](https://arxiv.org/abs/2105.00636))_

[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/learning-to-drive-from-a-world-on-rails/autonomous-driving-on-carla-leaderboard)](https://paperswithcode.com/sota/autonomous-driving-on-carla-leaderboard?p=learning-to-drive-from-a-world-on-rails)

This repo contains code for our paper [Learning to drive from a world on rails](https://arxiv.org/abs/2105.00636).

ProcGen code coming soon.

## Reference
If you find our repo or paper useful, please cite us as
```
@inproceedings{chen2021learning,
  title={Learning to drive from a world on rails},
  author={Chen, Dian and Koltun, Vladlen and Kr{\"a}henb{\"u}hl, Philipp},
  booktitle={arXiv preprint arXiv:2105.00636},
  year={2021}
}
```

## Updates
* Checkout our [website](https://dotchen.github.io/world_on_rails/) for demo videos!

## Getting Started
* To run CARLA and train the models, make sure you are using a machine with **at least** a mid-end GPU.
* Please follow [INSTALL.md](docs/INSTALL.md) to setup the environment.

## Training

* Please refer to [RAILS.md](docs/RAILS.md) on how to train our _World-on-Rails_ agent.
* Please refer to [LBC.md](docs/LBC.md) on how to train the _LBC_ agent.

## Evaluation

**If you evaluating the pretrained weights, make sure you are launching CARLA with `-vulkan`!**

### Leaderboard routes
```bash
python evaluate.py --agent-config=[PATH TO CONFIG]
```

### NoCrash routes
```bash
python evaluate_nocrash.py --town={Town01,Town02} --weather={train, test} --agent-config=[PATH TO CONFIG] --resume
```
* Use defaults for _RAILS_, and `--agent=autoagents/lbc_agent` for _LBC_.
* To print a readable table, use 
```bash
python -m scripts.view_nocrash_results [PATH TO CONFIG.YAML]
```

### Pretrained weights
* [Leaderboard models](https://utexas.box.com/s/8lcl7istkr23dtjqqiyu0v8is7ha5u2r)
* [NoCrash models](https://utexas.box.com/s/54m24gz5xwy1oagsqmgosch7pq561h2e)

## Dataset
We also release the data we trained for the leaderboard. 
Checkout [DATASET.md](docs/DATASET.md) for more details.

## Acknowledgements
The `leaderboard` codes are built from the original [leaderboard](https://github.com/carla-simulator/leaderboard.git) repo.
The `scenariorunner` codes are from the original [scenario_runner](https://github.com/carla-simulator/scenario_runner.git) repo.
The `waypointer.py` GPS coordinate conversion codes are build from Marin Toromanoff's leadeboard submission.

## License
This repo is released under the MIT License (please refer to the LICENSE file for details). The [leaderboard](https://github.com/carla-simulator/leaderboard.git) repo which our `leaderboard` folder builds upon is under the MIT License.