Robot Manipulation

Physical Intelligence, Bo Ai, Ali Amin et al.

$π_{0.7}$ delivers emergent, zero-shot robotic capabilities via a steerable foundation model, enabling complex multi-stage tasks in unseen environments—transforming how robots generalize across tasks and embodiments in real-world settings.

Ruihang Xu, Dewei Zhou, Xiaolong Shen et al.

Adds 3D geometry and physical constraints to image editing, plus a new benchmark, making object manipulation edits far more reliable for world-model, simulation, and synthetic-data workflows.

Songyuan Yang, Huibin Tan, Kailun Yang et al.

Programming domestic robots is often too complex for non-experts. This system allows users to instruct robots by simply sketching on a camera feed, removing the need for coding or pre-existing maps and making home robotics accessible to everyone.