PerchRL: Vision-Based Agile Perching on Inclined Platforms under Rapid and Irregular Motion
2026-06-02 • Robotics
RoboticsMachine Learning
AI summaryⓘ
The authors developed a system called PerchRL that helps small flying robots (quadrotors) land on moving slanted surfaces using cameras. They first train the system with simple data about the robot's position, then fine-tune it using actual camera images while handling times when the robot can't see well. To make the system work well with different motions, they use varied training paths and methods to understand past movements. Tests in simulations and real flights show their method is stable and works quickly on different types of quadrotors.
quadrotorperchingreinforcement learningvision-based controlfield of viewstate-based pre-trainingfine-tuningplatform trajectoryactive perceptionrobotic landing
Authors
Zihong Lu, Zongzhuo Liu, Huaxu Li, Jinqiang Cui, Jie Mei, Youmin Gong, U Kei Cheang, Boyu Zhou
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.