WAND: Learning Robust Navigation under Complex Wind Disturbances and Dense Obstacles for Quadrotors

2026-10-08

AI Summary

This study addresses the navigation challenge of quadrotors inferring time-varying disturbances under strong winds and dense obstacles by proposing the WAND framework. It employs temporal convolutional networks to estimate wind-induced acceleration and introduces a novel dual-path disturbance estimation mechanism, utilizing estimates simultaneously as policy conditioning inputs and low-level feedforward compensation. Coupled with a zero-initialized residual module for the navigation policy, this design enhances robustness in partially observable environments. Experimental results demonstrate an average 8.3% improvement in simulation success rates alongside wind-direction-adaptive trajectories. Furthermore, eighteen out of twenty indoor real-world flights succeeded, validating the feasibility of real-time onboard deployment.

Abstract

Robust navigation in cluttered environments remains a fundamental challenge for quadrotors, particularly when strong wind disturbances arise, which perturb vehicle dynamics, limit control authority, and substantially increase collision risk. Existing learning-based navigation policies typically rely on obstacle perception and proprioceptive observations, requiring the policy to infer time-varying disturbance effects implicitly and thereby limiting robustness under partial observability. This paper proposes WAND (Wind-Aware Navigation with Disturbance Estimation), a reinforcement learning framework for navigation under time-varying wind disturbances in dense obstacle fields. Specifically, WAND estimates wind-induced disturbance acceleration from historical proprioceptive states using a Temporal Convolutional Network (TCN). This estimation is integrated into the policy via a zero-initialized residual module, \emph{WindAdapter}, while simultaneously providing feedforward compensation for low-level control. The dual use of the estimate couples disturbance-conditioned navigation with feedforward disturbance rejection. Across 12 wind-disturbed simulation settings, WAND improved the observed success rate by 8.3 percentage points on average relative to feedforward compensation alone. Controlled opposite-crosswind experiments further showed wind-direction-dependent trajectory adaptation. In indoor fan-induced flight tests, WAND succeeded in 18 of 20 trials, demonstrating the feasibility of real-time onboard navigation.

Problem

• Quadrotor navigation
• Wind disturbance
• Dense obstacles
• Robust navigation
• Reinforcement learning

Innovation

• Reinforcement Learning
• Disturbance Estimation
• Temporal Convolutional Network
• Quadrotor Navigation
• Wind Disturbance Rejection