OpenHUTB

The OpenHUTB (Open source Human-vehicle Twin Build) is a photorealistic-quality physics simulator for studying human and vehicles (humanoid, autonomous vehicles, drones, and submersibles), designed to support the development, training, and validation of robotic algorithms. In addition to open-source code, it provides freely usable open digital assets (town layouts, buildings, vehicles, pedestrians, props, etc.) created for this purpose. The simulation platform supports standard specifications for VR cockpit, unified air-ground, submersibles, sensors, data synthesis, traffic management, multiphysics simulation, pedestrian navigation, and Python interfaces. For detailed information, please refer to the documentation.

English | 简体中文   |   🌐 Project Home  |  📖 Docs

Usage example

1. Run the following command in the terminal to download the simulator:

   # Download command for Windows; run CarlaUE4.exe to launch the simulator.
   curl -fsSL -o dl.bat https://openhutb.github.io/dl.bat && dl.bat
   
   # Download command for Ubuntu; run CarlaUE4.sh to launch the simulator.
   curl -fsSL -o dl.sh https://openhutb.github.io/dl.sh && chmod +x dl.sh && ./dl.sh
   

2. Navigate to the generated directory hutb/PythonAPI/carla/dist/, and use pip install hutb-*.whl to install the toolkit for your specific Python version (supports Python 3.7-3.14). Run the following script to generate vehicles and pedestrians in the scene:

   python PythonAPI/examples/generate_traffic.py
   

   Manual pedestrian control (this example uses pygame for the display window and keyboard input, so install it before the first run):

   pip install pygame
   python PythonAPI/examples/manual_control.py --filter walker.pedestrian.*
   

   Use config.py to switch to VR cockpit mode , and use the Logitech steering wheel or keyboard keys W, A, S, D, etc. for control; Z is reverse gear.

   python PythonAPI/util/config.py --map Town10HD?GAME=VR
   

Source code compilation

# Clone repository
git clone https://github.com/OpenHUTB/hutb.git
# Launch the editor
setup.bat -l
# Package
setup.bat -p
# Ubuntu
./setup.sh -l
./setup.sh -p

Please note that the hutb branch contains the latest version, as well as the latest fixes and features. Alternatively, follow the instructions in How to Build on Windows (Chinese Guide) and How to Build on Linux.

[!NOTE] Artists can skip compiling and directly download the software/hutb/hutb_editor.zip file from the link, extract it, and double-click launch_hutb_editor.bat to launch the Unreal Editor with the plugin.

Hardware and software requirements

• Processor: Intel i7 gen 9th - 11th / Intel i9 gen 9th - 11th / AMD ryzen 7 / AMD ryzen 9
• Memory: +16 GB
• Graphics card: NVIDIA RTX 2070 or higher
• OS: Windows 10+、Ubuntu 18.04+、MacOS 12+。

Ecosystem

Repositories related to the simulation platform:

• Related applications include: perception 、planning 、control 、end-to-end 、large language models 、pedestrians 、agents 、explainability ,etc.
• Autonomous Driving Leaderboard: An autonomous platform used to validate autonomous driving technology stacks
• Nvidia ecosystem: SimReady 、Neural rendering 、Cosmos Cosmos Transfer, etc.
• Scenario Runner: The engine that executes traffic scenarios in the simulator
• ROS-bridge: Interface between simulator and ROS
• Driving Benchmark: Benchmarking tools for autonomous driving tasks
• AutoWare AV stack: Bridge connecting the AutoWare AV stack and Carla
• Apollo Bridge: A data and control bridge for the communication between the latest version of Apollo and Carla
• Map Editor: A standalone GUI application that enhances RoadRunner maps with traffic light and sign information
• Reinforcement Learning: Code for various reinforcement learning models

Others

In addition to the documentation, supplementary content has been created for users. This is a good way to cover different topics, such as detailed explanations of specific modules, the latest improvements to features, future work, and so on.

• Conventional
  • Artistic Improvements: Environment and Rendering — Video | PPT
  • Core implementation: synchronization, snapshots, and landmarks — Video | PPT
  • Data Ingestion — Video | PPT
  • Pedestrians and their realization — Video | PPT
  • Sensors in the simulator — Video | PPT
• Modules
  • Mujoco plugin for simulator — mujoco_plugin
  • Drone plugin for simulator — air
  • VR Driving — interbehavior
  • Traffic Manager Improvements — Video | PPT
  • Integration of automotive software with ROS — Video | PPT
  • ScenarioRunner Introduction — Video | PPT
  • OpenSCENARIO Support — PPT
  • Biomechanics of human movement — Video | PPT | Video
• Features
  • Co-simulation with SUMO and PTV Vissim — Video | PPT
  • Integration of RSS-lib — PPT
  • External sensor interface（External Sensor Interface，ESI） — Video | PPT
  • OpenDRIVE Standalone Mode — Video | PPT

References and Licenses

This project stands on the shoulders of giants, and we sincerely thank the developers of the following open-source projects: Carla (MIT License), AirSim (MIT License), DReyeVR (MIT License), CarlaAir (MIT License), HoloOcean (MIT License), and OpenSim (Apache License).

Related assets are licensed under the CC-BY license. Other related assets (including map scenes of Hunan University of Technology and Business, Changsha CEC Software Park, Chinese Vehicles, etc.) and code are open source under the MIT license.