Skip to content

Latest commit

 

History

History
132 lines (93 loc) · 4.62 KB

README.md

File metadata and controls

132 lines (93 loc) · 4.62 KB

ScenarioMax

Python 3.10+ License: MIT

ScenarioMax is an extension to ScenarioNet that transforms various autonomous driving datasets into standardized formats. Like ScenarioNet, it first converts different datasets (Waymo, nuPlan, nuScenes) to a unified pickle format. ScenarioMax then extends this process with additional pipelines to convert this unified data into formats compatible with Waymax, V-Max, and GPUDrive.

Features

  • Data Unification Pipeline: Similar to ScenarioNet, converts raw datasets to a standardized pickle format with type corrections and data reordering
  • Multiple Output Formats:
    • TFRecord/TFExample: Generate optimized TFRecord files for Waymax and V-Max
    • JSON: Create compatible output for GPUDrive
  • SDC Path Support: Add and evaluate multiple potential trajectories for self-driving vehicles
  • Extensible Architecture: Easily add support for new datasets or output formats

Data Processing Pipeline

ScenarioMax provides a complete data transformation workflow:

  1. Raw to Unified (raw_to_unified):

    • Converts various datasets into a standardized pickle format
    • Applies type corrections and data reordering to ensure compatibility
    • Similar to ScenarioNet but with adjustments for downstream compatibility

  2. Unified to Target Format:

    • TFExample (unified_to_tfexample): Converts unified data to TFRecord format for Waymax/V-Max
    • GPUDrive (unified_to_gpudrive): Converts unified data to JSON format for GPUDrive

This two-stage pipeline allows for flexible processing of autonomous driving datasets across different simulation platforms.

Supported Datasets

Dataset Version Link Status
Waymo Open Motion Dataset v1.3.0 Site ✅ Full Support
nuPlan v1.1 Site ✅ Full Support
nuScenes v1.0 Site ✅ Full Support
Argoverse v2.0 Site 🚧 WIP

For dataset setup, you can see the complete ScenarioNet documention.

Quick Start Guide

Prerequisites

  • Python 3.10 or newer
  • Access to at least one supported dataset (Waymo, nuPlan, or nuScenes)
  • Sufficient disk space for dataset processing

Installation

Basic Installation

# Clone the repository
git clone https://github.com/yourusername/scenariomax.git
cd scenariomax

# Install the package
pip install -e .

With nuPlan Support

If you need to work with nuPlan data:

pip install -e devkit/nuplan-devkit
pip install -r devkit/nuplan-devkit/requirements.txt

Basic Usage

Converting Waymo Open Motion Dataset

python scenariomax/convert_dataset.py \
  --waymo_src /path/to/waymo/data \
  --dst /path/to/output/directory \
  --target_format tfexample \
  --num_workers 8 \
  --tfrecord_name training

Converting nuScenes Dataset

python scenariomax/convert_dataset.py \
  --nuscenes_src /path/to/nuscenes/data \
  --dst /path/to/output/directory \
  --split v1.0-trainval \
  --target_format tfexample \
  --num_workers 8 \
  --tfrecord_name training

Converting to GPUDrive Format

python scenariomax/convert_dataset.py \
  --waymo_src /path/to/waymo/data \
  --dst /path/to/output/directory \
  --target_format gpudrive \
  --num_workers 8

Understanding the Output

After running the conversion process, you'll have:

  • For tfexample format: TFRecord files compatible with Waymax and V-Max
  • For gpudrive format: JSON files compatible with GPUDrive
  • For pickle format: Standardized pickle files with unified scenario data. Pickle can be converted to a tfexample or a gpudrive afterwards.

Documentation

Release Notes

Version 1.0

  • Initial implementation with support for Waymo, nuPlan, and nuScenes
  • TFRecord conversion pipeline for efficient data handling
  • SDC paths support
  • GPUDrive format support

License

This project is licensed under the MIT License - see the LICENSE file for details.