tianlezeng/CarlaAIr-v0.1.7

CARLA-Air: Fly Drones Inside a CARLA World

A Unified Infrastructure for Air-Ground Embodied Intelligence

Overview

CARLA-Air is an open-source simulation infrastructure that unifies high-fidelity urban driving (CARLA) and physics-accurate multirotor flight (AirSim) within a single Unreal Engine process. It provides a practical foundation for air-ground embodied intelligence research, including:

• Air-ground cooperation (e.g., precision landing on moving vehicles)
• Embodied navigation & VLN/VLA (vision-language navigation with aerial + ground views)
• Multi-modal perception & dataset collection (12-stream synchronized capture)
• RL-based policy training (stable closed-loop air-ground interaction)

Key Features

Single-Process Integration	CARLA + AirSim in one UE4 process. No bridge, no latency. Only 3 upstream files modified (~35 lines).
Dual Python API	carla.Client (port 2000) + airsim.MultirotorClient (port 41451) in one script. 89/89 CARLA API tests passing.
Up to 18 Sensor Modalities	RGB, Depth, Semantic Seg, LiDAR, Radar, IMU, GNSS, and more — all frame-aligned across air and ground.
~20 FPS Joint Workloads	19.8 ± 1.1 FPS with vehicles + drone + 8 sensors. Communication overhead < 0.5 ms.
3-Hour Stability	357 spawn/destroy cycles, zero crashes, zero memory accumulation.
63 ROS 2 Topics	Full ROS 2 support across both simulation backends.

Download

This repository contains the prebuilt binary release of CARLA-Air v0.1.7 for Linux x86_64.

Files

File	Size	Description
CarlaAir-v0.1.7.zip	~6.4 GB	Complete binary release (Ubuntu 20.04 / 22.04)

Quick Start

# 1. Download and extract
# Download CarlaAir-v0.1.7.zip from the Files tab above
unzip CarlaAir-v0.1.7.zip
cd CarlaAir-v0.1.7

# 2. One-click environment setup
bash env_setup/setup_env.sh
conda activate carlaAir
bash env_setup/test_env.sh

# 3. Launch the simulator
./CarlaAir.sh Town10HD

# 4. Run the showcase (in another terminal)
conda activate carlaAir
python3 examples/quick_start_showcase.py

System Requirements

• OS: Ubuntu 20.04 / 22.04
• GPU: NVIDIA GPU with Vulkan support (tested on RTX A4000)
• Python: 3.8+
• Software: CARLA 0.9.16, AirSim 1.8.1, Unreal Engine 4.26

Links

• GitHub: louiszengCN/CarlaAir
• Paper: Technical Report (PDF)
• Baidu Pan: Download (提取码: d5ai)

Citation

@article{zeng2026carlaair,
  title   = {CARLA-Air: Fly Drones Inside a CARLA World --
             A Unified Infrastructure for Air-Ground Embodied Intelligence},
  author  = {Zeng, Tianle and Chen, Hanxuan and Wen, Yanci and Zhang, Hong},
  journal = {arXiv preprint arXiv:2603.28032},
  year    = {2026}
}

License

CARLA-Air specific code is distributed under the MIT License. CARLA specific assets are distributed under the CC-BY License.