Learning Visuomotor Quadrotor Flight Through Dynamic Narrow Gaps via Differentiable Simulation
作者: Riccardo Bianco, Yunfan Ren, Jiaxu Xing, Davide Scaramuzza
分类: cs.RO
发布日期: 2026-10-27
期刊: Conference on Robot Learning, 2026
💡 一句话要点
提出可微仿真方法以解决四旋翼穿越动态狭窄间隙问题
🎯 匹配领域: 支柱二:RL算法与架构 (RL & Architecture) 支柱八:物理动画 (Physics-based Animation)
关键词: 四旋翼飞行 动态间隙 可微仿真 控制策略 视觉导航 自主学习 路径优化
📋 核心要点
- 现有方法依赖于特权状态、教师或规划器,难以直接从零开始学习动态狭窄间隙穿越。
- 提出可微的动态间隙公式,通过有符号距离将穿越事件转化为路径目标,简化了控制策略的设计。
- 在合成基准测试中,路径梯度显著优于得分函数梯度,模拟中成功率达到99.3%,实际飞行成功率为80%。
📝 摘要(中文)
穿越动态狭窄间隙对四旋翼飞行提出了严峻挑战,要求厘米级的空间对齐和精确的时间执行。本文提出了一种控制策略,直接基于机载视觉和惯性测量输出推力和姿态命令。通过引入可微的动态间隙公式,将稀疏的穿越事件转化为密集的路径目标,避免了依赖教师或规划器的传统方法。实验结果表明,该方法在合成和真实环境中均表现出色,模拟中成功率达到99.3%,实际飞行中成功率为80%。
🔬 方法详解
问题定义:本文旨在解决四旋翼穿越动态狭窄间隙的控制问题。现有方法通常依赖于特权状态或教师指导,难以实现自主学习和适应性。
核心思路:提出了一种可微的动态间隙公式,利用有符号距离来引导四旋翼的对齐和中心化,避免了传统方法的局限性。
技术框架:整体架构包括从随机初始化开始的视觉特征到命令的策略训练,采用单阶段训练,无需教师或规划器。主要模块包括视觉输入处理、控制命令生成和路径优化。
关键创新:最重要的创新在于将稀疏的穿越事件转化为密集的路径目标,使得学习过程更加高效和稳定。与现有方法相比,减少了对外部信息的依赖。
关键设计:在损失函数中引入了平滑碰撞项和对齐项,确保了控制策略的精确性。网络结构设计上采用了适应性的卷积神经网络,以处理复杂的视觉输入。
🖼️ 关键图片
📊 实验亮点
实验结果显示,在合成环境中,路径梯度方法的成功率达到99.3%,而传统得分函数梯度方法效果较差。实际飞行实验中,四旋翼在多种初始条件下成功穿越动态狭窄间隙的成功率为80%,验证了方法的有效性与鲁棒性。
🎯 应用场景
该研究具有广泛的应用潜力,尤其是在无人机自主飞行、搜索与救援、物流运输等领域。通过提高四旋翼在动态环境中的适应能力,能够实现更高效的任务执行和安全性,未来可能推动无人机技术的进一步发展与普及。
📄 摘要(原文)
Threading a quadrotor through a moving narrow gap is a demanding test of agile flight, requiring centimeter-level spatial alignment and precise temporal execution. We address this challenge using a control policy that operates directly on onboard visual and inertial measurements to output collective thrust and body-rate commands. The task is difficult because success lies on a thin feasible manifold: small errors in timing, position, attitude, or visual observability cause collision. Prior planning and learning methods rely on privileged state, teachers, planners, or distillation; learning dynamic narrow-gap traversal directly, from scratch, remains open. We introduce a differentiable moving-gap formulation that gates centering, alignment, and smooth collision terms by signed distance to a time-varying gap plane, turning a sparse traversal event into a dense pathwise objective. We integrate this formulation into a visual-feature-to-command policy trained in one stage from random initialization, without a teacher, planner, or distillation. We validate our approach in both synthetic and real-world experiments. In controlled benchmarks, we demonstrate that pathwise gradients significantly outperform score-function gradients for precision-critical tasks, yielding a 99.3% success rate in simulation after only a few minutes of training on a single GPU. Furthermore, the policy transfers directly to a real quadrotor, successfully crossing a dynamic narrow gap from varied initial conditions using onboard sensing and inertial measurements, achieving an 80% success rate over 10 physical trials.