Vector Field Augmented Differentiable Policy Learning for Vision-Based Drone Racing
This work is significant for researchers and practitioners in autonomous drone racing, providing a more efficient and robust method for training agile drone policies.
The paper addresses the challenge of autonomous drone racing by proposing DiffRacing, a novel vector field-augmented differentiable policy learning framework. DiffRacing integrates differentiable losses and vector fields to provide continuous gradient signals, balancing obstacle avoidance and high-speed gate traversal, and achieves superior sample efficiency, faster convergence, and robust flight performance in both simulation and real-world experiments.
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.