Wenxian Yu

KLTNet achieves superior tracking accuracy and robustness in low-texture environments, outperforming classical methods while ensuring real-time processing capabilities.

Achieving superior text-image alignment, AnchorSteer outperforms traditional methods by actively correcting semantic errors during image generation.

VOP-Nav achieves a remarkable balance of speed and safety, outperforming all baseline methods in navigating crowded environments without explicit obstacle tracking.

Zero-shot sim-to-real transfer for quadrotor control is now systematically achievable, thanks to a unified framework that integrates differentiable physics learning with training, validation, and real-world deployment.

Ditch the point-mass abstraction: this RL policy directly controls quadrotor body rates from depth images, achieving unprecedented speed and stability in complex environments.