DiffAero: A GPU-Accelerated Differentiable Simulation Framework for Efficient Quadrotor Policy Learning

2025/09/12 by Xinhong Zhang, Zhang, Xinhong, Runqing Wang +11
Computer Science · Physics and Astronomy · #Advanced Neural Network Applications #Adversarial Robustness in Machine Learning #FOS: Computer and information sciences #Model Reduction and Neural Networks #Robotics (cs.RO)

paper · pdf · doi:10.48550/arxiv.2509.10247

openalex publication_date 2025/09/12 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

This letter introduces DiffAero, a lightweight, GPU-accelerated, and fully differentiable simulation framework designed for efficient quadrotor control policy learning. DiffAero supports both environment-level and agent-level parallelism and integrates multiple dynamics models, customizable sensor stacks (IMU, depth camera, and LiDAR), and diverse flight tasks within a unified, GPU-native training interface. By fully parallelizing both physics and rendering on the GPU, DiffAero eliminates CPU-GPU data transfer bottlenecks and delivers orders-of-magnitude improvements in simulation throughput. In contrast to existing simulators, DiffAero not only provides high-performance simulation but also serves as a research platform for exploring differentiable and hybrid learning algorithms. Extensive benchmarks and real-world flight experiments demonstrate that DiffAero and hybrid learning algorithms combined can learn robust flight policies in hours on consumer-grade hardware. The code is available at https://github.com/flyingbitac/diffaero.

Citations

• A General Infrastructure and Workflow for Quadrotor Deep Reinforcement Learning and Reality Deployment
• Aerial Gym Simulator: A Framework for Highly Parallelized Simulation of Aerial Robots
• What Matters in Learning A Zero-Shot Sim-to-Real RL Policy for Quadrotor Control? A Comprehensive Study
• Seeing Through Pixel Motion: Learning Obstacle Avoidance from Optical Flow with One Camera
• Learning Quadrotor Control From Visual Features Using Differentiable Simulation
• VisFly: An Efficient and Versatile Simulator for Training Vision-based Flight
• Back to Newton's Laws: Learning Vision-based Agile Flight via Differentiable Physics
• CGD: Constraint-Guided Diffusion Policies for UAV Trajectory Planning
• STORM: Efficient Stochastic Transformer based World Models for Reinforcement Learning
• OmniDrones: An Efficient and Flexible Platform for Reinforcement Learning in Drone Control
• Champion-level drone racing using deep reinforcement learning
• Efficient and Robust Time-Optimal Trajectory Planning and Control for Agile Quadrotor Flight
• Accelerated Policy Learning with Parallel Differentiable Simulation
• Learning to Walk in Minutes Using Massively Parallel Deep Reinforcement\n Learning
• Isaac Gym: High Performance GPU-Based Physics Simulation For Robot Learning
• Brax -- A Differentiable Physics Engine for Large Scale Rigid Body Simulation
• Robust Navigation for Racing Drones based on Imitation Learning and Modularization
• Learning to Fly -- a Gym Environment with PyBullet Physics for\n Reinforcement Learning of Multi-agent Quadcopter Control
• The Surprising Effectiveness of PPO in Cooperative, Multi-Agent Games
• Flightmare: A Flexible Quadrotor Simulator
• PyTorch: An Imperative Style, High-Performance Deep Learning Library
• Proximal Policy Optimization Algorithms
• AirSim: High-Fidelity Visual and Physical Simulation for Autonomous Vehicles

Related

• Flightmare: A Flexible Quadrotor Simulator
• Aerial Gym Simulator: A Framework for Highly Parallelized Simulation of Aerial Robots
• VisFly: An Efficient and Versatile Simulator for Training Vision-based Flight
• Learning to Fly in Seconds
• What Matters in Learning A Zero-Shot Sim-to-Real RL Policy for Quadrotor Control? A Comprehensive Study