基于向量场的可微政策学习用于基于视觉的无人机赛车
Vector Field Augmented Differentiable Policy Learning for Vision-Based Drone Racing
AI总结：
本文提出DiffRacing框架，通过整合向量场和可微损失提升无人机赛车的样本效率与飞行性能。
摘要：
在复杂环境中实现自主无人机赛车需要敏捷且高速的飞行，同时保持可靠的障碍物避让。基于可微物理的政策学习最近在各种任务中展示了高样本效率和显著的性能，包括敏捷无人机飞行和四足运动。然而，将此类方法应用于无人机赛车仍然具有挑战性，因为关键目标如通过门的性能本质上难以表示为平滑、可微的损失。为了解决这些挑战，我们提出了DiffRacing，一种新颖的基于向量场的可微政策学习框架。DiffRacing将可微损失和向量场整合到训练过程中，以提供连续且稳定的梯度信号，平衡障碍物避让和高速通过门。此外，一个可微的Delta动作模型补偿了动态不匹配，使模拟到现实的转移变得高效，而无需显式的系统识别。广泛的模拟和现实世界实验表明，DiffRacing实现了更高的样本效率、更快的收敛速度和鲁棒的飞行性能，从而证明了向量场可以利用任务特定的几何先验来增强传统基于梯度的政策学习。
英文摘要：
Autonomous drone racing in complex environments requires agile, high-speed flight while maintaining reliable obstacle avoidance. Differentiable-physics-based policy learning has recently demonstrated high sample efficiency and remarkable performance across various tasks, including agile drone flight and quadruped locomotion. However, applying such methods to drone racing remains difficult, as key objective like gate traversal are inherently hard to express as smooth, differentiable losses. To address these challenges, we propose DiffRacing, a novel vector field-augmented differentiable policy learning framework. DiffRacing integrates differentiable losses and vector fields into the training process to provide continuous and stable gradient signals, balancing obstacle avoidance and high-speed gate traversal. In addition, a differentiable Delta Action Model compensates for dynamics mismatch, enabling efficient sim-to-real transfer without explicit system identification. Extensive simulation and real-world experiments demonstrate that DiffRacing achieves superior sample efficiency, faster convergence, and robust flight performance, thereby demonstrating that vector fields can augment traditional gradient-based policy learning with a task-specific geometric prior.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
