📖 Table of Contents

• 🌟 Project Overview
• ⚡ Performance
• 🔧 Supported Hardware
• 🎨 Visualization
• 🚀 Getting Started
• 📚 Integration Guide
• 🤝 Contributing

📋 Project Status

⚡ Performance Highlights

• 🔥 Ultra-Fast Rendering: 250ms per step for 4,096 environments with 20,000 rays each (RTX 4090)
• 🎯 Multi-Sensor Support: 11+ lidar sensor types including Livox and Velodyne series
• 🌐 Multi-Platform: Supports IsaacGym, Genesis, Mujoco, and Isaac Sim
• ⚡ GPU Acceleration: CUDA-optimized ray tracing with Warp backend

✨ Key Features

🔧 Supported Hardware

Lidar Sensors

Simulation Platforms

• IsaacGym - NVIDIA's physics simulation platform
• Genesis - High-performance physics engine
• IsaacLab - Next-generation robotics simulation platform
• Mujoco - Advanced physics simulation
• Isaac Sim - Omniverse-based robotics simulation

🎨 Visualization Examples

Real-time Lidar Simulation

💡 Development Status

📢 Important Note: This project is under active development. While the LidarSensor module is fully functional and optimized, we're continuously improving documentation and code structure. For any issues or questions, please open an issue.

🚀 Getting Started

📦 Installation

Prerequisites

• Python 3.8+
• CUDA 11.0+ (for GPU acceleration)
• One of: IsaacGym, Genesis, Mujoco, or Isaac Sim

Quick Install

# Install core dependencies
pip install warp-lang[extras] taichi

# Install LidarSensor
cd LidarSensor
pip install -e .

Optional Dependencies

# For ROS integration (optional)
source /opt/ros/humble/setup.bash

# For advanced visualization (optional)  
pip install matplotlib open3d

🎯 Quick Usage

1. Basic Example - IsaacGym

# Generate self-occlusion mesh (first time only)
cd LidarSensor/resources/robots/g1_29/
python process_body_mesh.py

# Run example with Unitree G1
cd LidarSensor/example/isaacgym
python unitree_g1.py

2. ROS Integration Example

# Start ROS visualization
source /opt/ros/humble/setup.bash
/usr/bin/python3 LidarSensor/LidarSensor/sensor_pattern/sensor_lidar/lidar_vis_ros2.py

3. Custom Configuration

from LidarSensor.sensor_config.lidar_sensor_config import LidarConfig, LidarType

# Create custom sensor configuration
config = LidarConfig(
    sensor_type=LidarType.MID360,
    max_range=30.0,
    enable_sensor_noise=False
)

🚀 Platform-Specific Integration Guides

Choose your simulation platform for detailed installation and usage instructions:

📦 IsaacLab - Recommended for large-scale RL training

Native LiDAR integration with 7+ Livox sensor types and optimized performance

Key Features:

• ✅ Native LidarSensor class integration
• ✅ Realistic Livox patterns (.npy files)
• ✅ Optimized for 1000+ environments
• ✅ Easy dataclass configuration

📂 Complete IsaacLab Guide

🎮 IsaacGym - For NVIDIA GPU-accelerated physics

Direct GPU ray-casting with Warp integration

Key Features:

• ✅ GPU-accelerated ray tracing
• ✅ Multiple sensor configurations
• ✅ Real-time visualization
• ✅ Flexible terrain integration

📂 Complete IsaacGym Guide

🌟 Genesis - For high-performance physics simulation

Modern physics engine with optimized LiDAR support

Key Features:

• ✅ High-performance physics
• ✅ Multiple robot platforms
• ✅ Realistic sensor modeling
• ✅ Cross-platform support

📂 Complete Genesis Guide

🤝 Contributing & Support

• 🐛 Issues: Report bugs or request features via GitHub Issues
• 📖 Documentation: Platform-specific guides in /example/ directories
• 🔬 Research: Cite our work if you use OmniPerception in research
• 💬 Discussions: Join our community for tips and collaboration

Cite

@article{wang2025omni,
  title={Omni-Perception: Omnidirectional Collision Avoidance for Legged Locomotion in Dynamic Environments},
  author={Wang, Zifan and Ma, Teli and Jia, Yufei and Yang, Xun and Zhou, Jiaming and Ouyang, Wenlong and Zhang, Qiang and Liang, Junwei},
  journal={arXiv preprint arXiv:2505.19214},
  year={2025}
}

Acknowledgments