PerchRL: Vision-Based Agile Perching on Inclined Platforms under Rapid and Irregular Motion

2026-06-02

AI Summary

This work addresses the challenge of enabling quadrotor drones to achieve stable vision-based perching on rapidly moving, irregularly tilted platforms—a task hindered by limited field of view and intermittent visual loss. To tackle this, the authors propose a two-stage reinforcement learning framework: an initial state-based pretraining phase followed by a vision-based fine-tuning stage that incorporates visibility-aware state augmentation and an active perception reward mechanism. By integrating randomized platform trajectories and temporal data augmentation, the approach significantly enhances policy generalization and robustness under visual interruptions. The method demonstrates high success rates, real-time performance, and stability in both simulation and real-world experiments, and has been successfully deployed across multiple quadrotor platforms, confirming its effectiveness and transferability.

Abstract

Autonomous vision-based perching of quadrotors on moving inclined platforms is critical for air-ground collaboration but remains challenging due to the limited field of view (FOV). In this paper, we propose PerchRL, a reinforcement learning (RL) framework for vision-based agile perching on inclined platforms under rapid and irregular motion. Specifically, we employ a two-stage learning strategy consisting of state-based pre-training followed by vision-based fine-tuning. To improve generalization across diverse platform motions, we employ randomized platform trajectories to prevent overfitting and temporal augmentation methods to capture latent motion patterns from historical observations. During vision-based fine-tuning, a hybrid learning framework consisting of visibility-aware state augmentation and active perception rewards is presented to improve robustness under intermittent visual loss. Extensive simulation and real-world experiments demonstrate the feasibility, stability, and real-time performance of PerchRL, while successful deployment across distinct quadrotor platforms further validates its adaptability. The source code will be released to benefit the community.

Problem

• vision-based perching
• inclined platforms
• quadrotors
• rapid and irregular motion
• limited field of view

Innovation

• reinforcement learning
• vision-based perching
• quadrotor
• active perception
• temporal augmentation