VisionFlow-PX4 Developer Manual

A customized PX4 Autopilot fork developed by WindyLab, integrating UAVs with robotic arms (Gamma series) for manipulation tasks in Gazebo simulation, featuring Prescribed Performance Guidance and Management Estimator (PreGME) control and ROS2 integration.

Core Features

• PreGME Controllers — Full replacement of standard mc_att_control and mc_pos_control with sliding-mode Prescribed Performance Control (PPC) algorithms, including centroid compensation and composite error state observer (CESO)
• Gamma Arm Integration — Tight coupling between PX4 flight control and Gamma-series robotic arm dynamics via the gamma_arm_dynamics library
• Rich Gazebo Simulation — Custom worlds, models, and plugins for indoor laboratory manipulation scenarios (landing boxes, VLA tasks, hardware-in-the-loop)
• ROS2 Ecosystem — Zenoh middleware, uXRCE-DDS client, and a complete ROS2 Humble Docker environment

Key Differences from Stock PX4

1. PreGME Controllers — Sliding-mode PPC replaces standard MC controllers (core research contribution)
2. Robotic Arm Integration — gamma_arm_dynamics bridges PX4 flight control with Gamma-series arm dynamics
3. Heavy Gazebo Simulation — Custom worlds, models, plugins for indoor lab manipulation
4. ROS2 Integration — Zenoh middleware, uXRCE-DDS, Gazebo-ROS bridge, complete ROS2 Humble Docker
5. Camera Feedback Pipeline — OAK-D and Intel RealSense support with geotagging
6. Differential Rover Support — Full rover control stack alongside quadcopter

Quick Navigation

Section	Content
Getting Started	Environment setup, Docker launch, native launch
System Architecture	Control stack, simulation stack, communication stack
Core Modules	PreGME controllers, arm integration, NN control, etc.
Simulation Assets	World scenes, models, asset usage guide
Toolchain	Docker workflow, flight review, control panels
Hardware Support	HKUST boards, supported boards list
Communication Protocols	Custom uORB messages, message reference
Development Guide	Build workflow, module development, contributing
References	Airframes, parameters, citations

Citation

If you use this codebase in your research, please cite the associated paper:

@article{ji2025pregme,
  title={PreGME: Prescribed Performance Control of Aerial Manipulators based on Variable-Gain ESO},
  author={Ji, Mengyu and Guo, Shiliang and Li, Zhengzhen and Shen, Jiahao and Cao, Huazi and Zhao, Shiyu},
  journal={arXiv preprint arXiv:2512.22957},
  year={2025}
}