// the find
robin-shaun/XTDrone
UAV Simulation Platform based on PX4, ROS and Gazebo
XTDrone is a UAV simulation platform built on top of PX4, ROS 1, and Gazebo, targeting researchers who need to validate algorithms (SLAM, motion planning, swarm coordination) before deploying to real hardware. It supports multirotors, fixed-wings, VTOLs, ground vehicles, and surface vessels in a single framework, backed by two published papers. The primary audience is academic — Chinese university robotics labs and competition participants.
1. Vehicle breadth is real: multirotors, fixed-wing, VTOL (quadplane, tailsitter, tiltrotor), UGV, USV, and aerial manipulators all in one place — most comparable platforms only do one category well. 2. The hierarchical multi-UAV architecture separates ground control, communication, and flight control cleanly, which matters when you're running 20+ simulated vehicles and need to debug message routing. 3. Contributor demos are checked in as working examples with launch files and world files, not just README screenshots — you can actually reproduce what the GIFs show. 4. A successor (XTDrone2 targeting ROS 2) already exists, meaning the PX4/ROS 1 stack isn't a dead end for the team.
1. ROS 1 (Melodic/Noetic) is EOL and the entire codebase is built around it — migrating your own algorithms to this platform now means inheriting a stack with no upstream security fixes and a 2025 sunset. The ROS 2 fork is a separate repo with its own maturity questions. 2. Documentation lives entirely on Yuque (a Chinese knowledge-base SaaS), not in the repo — if that service goes down or paywalls content, you have no fallback. The README is mostly GIFs and a citation request. 3. Build artifacts are committed to the repo (compiled CMakeCache, `.out` binaries, a zip file of devel/), which signals that installation is fragile enough that maintainers are shipping their build state rather than fixing the build reproducibility. 4. English documentation is a thin translation of the Chinese README; the actual setup guide, API docs, and troubleshooting are only in Chinese, which is a hard wall for non-Chinese-reading users.