uzh-rpg/flightmare

Health v2 · maintenance only

Flags: no_license

Adoption not part of the score

1406 stars · 401 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Flightmare is a modular quadrotor simulator from the UZH Robotics and Perception Group, composed of a decoupled Unity-based rendering engine and a flexible physics engine that can run independently. It offers a multi-modal sensor suite, parallel multi-agent simulation with an OpenAI Gym-style API, ROS integration, and VR headset support.

Use cases

• simulate quadrotor flight for reinforcement learning research
• train drone control policies with hundreds of parallel agents
• generate synthetic RGB, depth, and segmentation data for deep learning
• test visual-inertial odometry and SLAM algorithms
• develop and evaluate path-planning and model-based control for drones
• run human-robot interaction experiments in virtual reality

When to choose

• you need a flexible, modular drone simulator with decoupled rendering and physics
• you want OpenAI Gym-style environments for quadrotor RL
• you need photo-realistic sensor simulation (RGB, depth, segmentation, point clouds) for drones
• you work with ROS and want an alternative to Gazebo-based MAV simulators

When to avoid

• you need a simulator for vehicles other than quadrotors
• you require a actively developed project with frequent updates and support
• you need a turnkey simulator without Linux/ROS setup effort
• you need guaranteed licensing clarity beyond the MIT-labeled code

Facets

application · maturity maintenance

simulation machine-learning reinforcement-learning robotics image-processing graphics robotics simulation machine-learning reinforcement-learning autonomous-vehicles cpp python quadrotor drone-simulator unity-rendering physics-engine ros openai-gym visual-inertial-odometry virtual-reality multi-agent linux docker

Member repositories

For agents

markdown · JSON · MCP: product_card(name="uzh-rpg/flightmare")