Bridging Performance and Generalization in Reinforcement Learning for Agile Flight

• URL: http://arxiv.org/abs/2606.27348v1
• Date: Thu, 25 Jun 2026 17:54:36 GMT
• Title: Bridging Performance and Generalization in Reinforcement Learning for Agile Flight
• Abstract summary: We propose a framework for zero-shot generalization in agile flight for RL-based drone racing.<n>Our method achieves strong zero-shot performance across a wide range of unseen racetracks in the real world.
• Score: 26.465258512733126
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Autonomous drone racing is a fundamentally challenging regime for autonomous aerial robots, requiring time-optimal control while operating under persistent actuation saturation. While reinforcement learning (RL) has achieved human-level performance in this domain, current methods fail to generalize; policies trained on specific environments often crash immediately in unseen configurations. This failure reflects the intrinsic difficulty of zero-shot generalization in agile flight, arising from high-dimensional task variation and the tight coupling between safety and performance at high speeds. Existing approaches that improve generalization impose a substantial cost on flight speed: control policies must significantly degrade performance to achieve even modest levels of generalization. In this work, we propose a framework for zero-shot generalization in agile flight for RL-based drone racing. By combining task-aware switching based on learning progress with a physically informed procedural track generator, the framework produces a fast and robust generalist policy without test-time adaptation. Our method achieves strong zero-shot performance across a wide range of unseen racetracks in the real world, demonstrating a 7.4x improvement in generalization over the state-of-the-art approaches, while maintaining competitive racing speeds. We validate our method's results in both simulation and real-world settings, including a challenging vision-based, end-to-end control setting that operates without explicit state estimation, where all prior approaches fail to generalize.

Related papers

• AgilePE: Autonomous UAV Pursuit-Evasion via Self-Play Reinforcement Learning [31.828745787124316]
  We present AgilePE, a complete system for autonomous UAV pursuit-evasion via self-play reinforcement learning.<n> AgilePE integrates agile low-level control, competitive policy optimization, and sim-to-real deployment in a unified framework.<n>For training, we use competitive self-play with Prioritized Fictitious Self-Play (PFSP) and a diversified opponent pool, enabling agents to improve against historical policies.<n>Real-world experiments reproduce pursuit-evasion tactics observed in simulation, including rapid dodging and flanking, and demonstrate interactive two-agent zero-shot deployment.
  arXiv  Detail & Related papers  (2026-08-14T09:41:16Z)
• Mind Your Steps: A General Learning Framework for Accurate Humanoid Foothold Tracking [78.30338456206984]
  We introduce a novel framework for training general-purpose 3D foothold-tracking policies.<n>By dynamically providing footstep support through a goal sampler, this method enables the learned policy to be agnostic to specific terrains.<n>Our new target representation effectively mitigates challenges arising in the real world, such as noisy and inaccurate pose estimation and foot contact estimation.
  arXiv  Detail & Related papers  (2026-06-06T16:44:56Z)
• Learning Reactive Dexterous Grasping via Hierarchical Task-Space RL Planning and Joint-Space QP Control [50.28263951510334]
  We propose a hybrid hierarchical control framework for reactive dexterous grasping.<n>The proposed framework explicitly decouples high-level spatial intent from low-level joint execution.<n>We extensively validate the proposed framework through a rigorous simulation-to-reality pipeline.
  arXiv  Detail & Related papers  (2026-05-05T04:49:38Z)
• Simple but Stable, Fast and Safe: Achieve End-to-end Control by High-Fidelity Differentiable Simulation [14.322775509774246]
  Obstacle avoidance is a fundamental vision-based task essential for enabling quadrotors to perform advanced applications.<n>We propose a novel end-to-end policy that directly maps depth images to low-level bodyrate commands by reinforcement learning.<n>Results demonstrate that the proposed approach achieves the highest success rate and the lowest jerk among state-of-the-art baselines.
  arXiv  Detail & Related papers  (2026-04-12T09:38:00Z)
• Agile Interception of a Flying Target using Competitive Reinforcement Learning [2.982218441172364]
  This article presents a solution to intercept an agile drone by another agile drone carrying a catching net.<n>We formulate the interception as a Competitive Reinforcement Learning problem, where the interceptor and the target drone are controlled by separate policies.<n>We introduce a high-fidelity simulation environment that integrates a realistic quadrotor dynamics model and a low-level control architecture implemented in JAX.
  arXiv  Detail & Related papers  (2026-03-17T09:11:23Z)
• Curriculum Reinforcement Learning for Quadrotor Racing with Random Obstacles [12.806994271928483]
  We propose a novel vision-based curriculum reinforcement learning framework for training a robust controller capable of addressing unseen obstacles in drone racing.<n>Our end-to-end control policy is implemented as a single network, allowing high-speed flight of quadrotors in environments with variable obstacles.
  arXiv  Detail & Related papers  (2026-02-27T13:56:59Z)
• Mean Flow Policy with Instantaneous Velocity Constraint for One-step Action Generation [65.13627721310613]
  Mean velocity policy (MVP) is a new generative policy function that models the mean velocity field to achieve the fastest one-step action generation.<n>MVP achieves state-of-the-art success rates across several challenging robotic manipulation tasks from Robomimic and OGBench.
  arXiv  Detail & Related papers  (2026-02-14T14:44:06Z)
• High-Speed Vision-Based Flight in Clutter with Safety-Shielded Reinforcement Learning [7.796497371549364]
  We propose an end-to-end RL framework augmented with model-based safety mechanisms.<n>During training, we design a physics-informed reward structure that provides global navigational guidance.<n>During deployment, we integrate a real-time safety filter that projects the policy outputs onto a provably safe set to enforce strict collision-avoidance constraints.
  arXiv  Detail & Related papers  (2026-02-09T13:47:02Z)
• Mastering Diverse, Unknown, and Cluttered Tracks for Robust Vision-Based Drone Racing [22.63301233637327]
  Most reinforcement learning methods for drone racing target fixed, obstacle-free tracks, leaving the generalization to unknown, cluttered environments largely unaddressed.<n>This challenge stems from the need to balance racing speed and collision avoidance, limited feasible space causing policy exploration trapped in local optima during training.<n>We propose a two-phase learning framework: an initial soft-collision training phase that preserves policy exploration for high-speed flight, followed by a hard-collision refinement phase that enforces robust obstacle avoidance.
  arXiv  Detail & Related papers  (2025-12-10T12:02:48Z)
• FASTer: Toward Efficient Autoregressive Vision Language Action Modeling via Neural Action Tokenization [61.10456021136654]
  We introduce FASTer, a unified framework for efficient and general robot learning.<n>FASTerVQ encodes action chunks as single-channel images, capturing global-temporal dependencies while maintaining a high compression ratio.<n>FASTerVLA builds on this tokenizer with block-wise autoregressive decoding and a lightweight action expert, achieving both faster inference and higher task performance.
  arXiv  Detail & Related papers  (2025-12-04T16:21:38Z)
• SpeedAug: Policy Acceleration via Tempo-Enriched Policy and RL Fine-Tuning [52.29534291796025]
  Reinforcement learning is a promising approach that adapts policies for faster execution without additional demonstration.<n>We propose SpeedAug, an RL-based policy acceleration framework that efficiently adapts pre-trained policies for faster task execution.
  arXiv  Detail & Related papers  (2025-11-24T04:25:47Z)
• Learned Controllers for Agile Quadrotors in Pursuit-Evasion Games [42.74003740156243]
  We address the problem of agile 1v1 quadrotor pursuit-evasion, where a pursuer and an evader learn to outmaneuver each other.<n>We propose an Asynchronous Multi-Stage Population-Based (AMSPB) algorithm to tackle these issues.<n>Within this framework, we train neural network controllers that output either velocity commands or body rates with collective thrust.
  arXiv  Detail & Related papers  (2025-06-03T13:19:23Z)
• RACER: Epistemic Risk-Sensitive RL Enables Fast Driving with Fewer Crashes [57.319845580050924]
  We propose a reinforcement learning framework that combines risk-sensitive control with an adaptive action space curriculum. We show that our algorithm is capable of learning high-speed policies for a real-world off-road driving task.
  arXiv  Detail & Related papers  (2024-05-07T23:32:36Z)
• Guided Constrained Policy Optimization for Dynamic Quadrupedal Robot Locomotion [78.46388769788405]
  We introduce guided constrained policy optimization (GCPO), an RL framework based upon our implementation of constrained policy optimization (CPPO) We show that guided constrained RL offers faster convergence close to the desired optimum resulting in an optimal, yet physically feasible, robotic control behavior without the need for precise reward function tuning.
  arXiv  Detail & Related papers  (2020-02-22T10:15:53Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.