Curriculum Reinforcement Learning for Quadrotor Racing with Random Obstacles

2026-02-27

AI Summary

This work proposes a vision-based curriculum reinforcement learning framework to address the perception and dynamic control challenges faced by quadrotor drones during high-speed racing through tracks with random obstacles. By integrating multi-stage curriculum learning, domain randomization, and a novel multi-scenario policy update strategy, the approach uniquely combines curriculum reinforcement learning with cross-scenario policy refinement to enable end-to-end training of a vision-driven controller that effectively balances the competing demands of obstacle avoidance and gate traversal. Experimental results demonstrate that the proposed method significantly improves lap completion rates and reduces lap times in both simulation and real-world environments, enhancing policy generalization and robustness, and thereby advancing the state of autonomous drone racing in dense, cluttered scenarios.

Technology Category

• Computer Vision: Learning & Optimization for CV
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Planning, Routing, and Scheduling: Replanning and Plan Repair

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Autonomous drone racing has attracted increasing interest as a research topic for exploring the limits of agile flight. However, existing studies primarily focus on obstacle-free racetracks, while the perception and dynamic challenges introduced by obstacles remain underexplored, often resulting in low success rates and limited robustness in real-world flight. To this end, we propose a novel vision-based curriculum reinforcement learning framework for training a robust controller capable of addressing unseen obstacles in drone racing. We combine multi-stage cu rriculum learning, domain randomization, and a multi-scene updating strategy to address the conflicting challenges of obstacle avoidance and gate traversal. Our end-to-end control policy is implemented as a single network, allowing high-speed flight of quadrotors in environments with variable obstacles. Both hardware-in-the-loop and real-world experiments demonstrate that our method achieves faster lap times and higher success rates than existing approaches, effectively advancing drone racing in obstacle-rich environments. The video and code are available at: https://github.com/SJTU-ViSYS-team/CRL-Drone-Racing.

Problem

• quadrotor racing
• obstacle avoidance
• robustness
• autonomous drone
• vision-based control

Innovation

• Curriculum Reinforcement Learning
• Domain Randomization
• Vision-based Control
• Quadrotor Racing
• Obstacle Avoidance