The field of robot control is witnessing a significant shift towards diffusion-based methods, which have shown impressive results in various tasks such as robotic manipulation, object pose estimation, and video-to-action control. These methods have been able to bridge the gap between high-level motion intent and low-level robot action, enabling more efficient and scalable robot learning. The use of diffusion models has also led to improved performance in tasks that require satisfaction of kinematic equality constraints. Notably, the integration of diffusion modeling with motion-centric representations has emerged as a strong baseline for robust robot learning. Noteworthy papers include: DAWN, which presents a unified diffusion-based framework for language-conditioned robotic manipulation. SCOPE, which introduces a diffusion-based category-level object pose estimation model that eliminates the need for discrete category labels. U-DiT Policy, which proposes a novel U-shaped Diffusion Transformer framework for robotic manipulation. PoseDiff, which presents a conditional diffusion model that unifies robot state estimation and control within a single framework. Act to See, See to Act, which introduces a unified representation learning that explicitly models a dynamic interplay between perception and action through probabilistic latent dynamics.