---
dataset_info:
  features:
  - name: run_id
    dtype: string
  - name: frame
    dtype: int32
  - name: timestamp
    dtype: float32
  - name: image_front
    dtype: image
  - name: image_front_left
    dtype: image
  - name: image_front_right
    dtype: image
  - name: image_rear
    dtype: image
  - name: location_x
    dtype: float32
  - name: location_y
    dtype: float32
  - name: location_z
    dtype: float32
  - name: rotation_pitch
    dtype: float32
  - name: rotation_yaw
    dtype: float32
  - name: rotation_roll
    dtype: float32
  - name: velocity_x
    dtype: float32
  - name: velocity_y
    dtype: float32
  - name: velocity_z
    dtype: float32
  - name: speed_kmh
    dtype: float32
  - name: throttle
    dtype: float32
  - name: steer
    dtype: float32
  - name: brake
    dtype: float32
  - name: nearby_vehicles_50m
    dtype: int32
  - name: total_npc_vehicles
    dtype: int32
  - name: total_npc_walkers
    dtype: int32
  - name: map_name
    dtype: string
  - name: weather_cloudiness
    dtype: float32
  - name: weather_precipitation
    dtype: float32
  - name: weather_fog_density
    dtype: float32
  - name: weather_sun_altitude
    dtype: float32
  - name: vehicles_spawned
    dtype: int32
  - name: walkers_spawned
    dtype: int32
  - name: duration_seconds
    dtype: int32
  splits:
  - name: train
    num_bytes: 155077419480.6
    num_examples: 56200
  - name: validation
    num_bytes: 14948709540
    num_examples: 4800
  - name: test
    num_bytes: 17602075134
    num_examples: 7200
  download_size: 189226141844
  dataset_size: 187628204154.6
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: validation
    path: data/validation-*
  - split: test
    path: data/test-*
license: mit
ta:
- vision-to-control
- imitation-learning
- autonomous-driving
- multimodal
- computer-vision
- reinforcement-learning
language:
- en
pretty_name: CARLA Autopilot Image Dataset
size_categories:
- 10K<n<100K
task_categories:
- image-feature-extraction
- any-to-any
- reinforcement-learning
---

# CARLA Autopilot Images Dataset

> **Note:** A newer, extended version of this dataset is available.  
> 🤗 [CARLA Autopilot Multimodal Dataset](https://huggingface.co/datasets/immanuelpeter/carla-autopilot-multimodal-dataset) 🤗  
> It includes **semantic segmentation, LiDAR, 2D bounding boxes, and additional environment metadata**.  
> Use it if your research requires multimodal signals beyond the RGB images and vehicle state/control data provided here.

This dataset contains autonomous driving data collected from CARLA simulator using autopilot.

## Dataset Structure

- **Train/val/test split is by run, not by frame, to ensure generalization**
- **Total train samples**: 56.2K
- **Total val samples**: 4.8K
- **Total test samples**: 7.2K
- **Runs processed**: 24

## Features

### Images
Multiple camera views are available depending on the run configuration:
- `image_front`: Front-facing camera view
- `image_rear`: Rear-facing camera view
- `image_front_left`: Front-left camera view
- `image_front_right`: Front-right camera view

### Vehicle State
- Position: `location_x`, `location_y`, `location_z`
- Orientation: `rotation_pitch`, `rotation_yaw`, `rotation_roll`
- Velocity: `velocity_x`, `velocity_y`, `velocity_z`, `speed_kmh`

### Vehicle Controls (Targets)
- `throttle`: Throttle input [0.0, 1.0]
- `steer`: Steering input [-1.0, 1.0]
- `brake`: Brake input [0.0, 1.0]

### Environment
- Traffic density information
- Weather conditions
- Map information

## Usage

```python
from datasets import load_dataset

dataset = load_dataset("immanuelpeter/carla-autopilot-images")
train_dataset = dataset["train"]
val_dataset = dataset["validation"]
test_dataset = dataset["test"]
```

## Citation

If you use this dataset, please cite the CARLA simulator:
```
@inproceedings{Dosovitskiy17,
  title = {CARLA: An Open Urban Driving Simulator},
  author = {Alexey Dosovitskiy and German Ros and Felipe Codevilla and Antonio Lopez and Vladlen Koltun},
  booktitle = {Proceedings of the 1st Annual Conference on Robot Learning},
  pages = {1--16},
  year = {2017}
}
```