VisFly: An Efficient and Versatile Simulator for Training Vision-based Flight

TL;DR

VisFly is a high-performance, versatile simulator for training vision-based quadrotor flight policies using reinforcement learning, supporting diverse environments and easy integration with learning algorithms.

Contribution

Introduces VisFly, a fast, user-friendly simulator with differentiable physics and scene dataset support, optimized for training vision-based flight policies.

Findings

Achieves over 10,000 fps rendering performance.

Supports diverse real-world environment datasets.

Provides reinforcement learning examples for flight tasks.

Abstract

We present VisFly, a quadrotor simulator designed to efficiently train vision-based flight policies using reinforcement learning algorithms. VisFly offers a user-friendly framework and interfaces, leveraging Habitat-Sim's rendering engines to achieve frame rates exceeding 10,000 frames per second for rendering motion and sensor data. The simulator incorporates differentiable physics and is seamlessly wrapped with the Gym environment, facilitating the straightforward implementation of various learning algorithms. It supports the directly importing open-source scene datasets compatible with Habitat-Sim, enabling training on diverse real-world environments simultaneously. To validate our simulator, we also make three reinforcement learning examples for typical flight tasks relying on visual observations. The simulator is now available at [https://github.com/SJTU-ViSYS-team/VisFly].