DreamVu Research on NVIDIA's Cosmos3 Overturns a Core Assumption in Robot Training — The Overlooked Camera Angle Beats More Data

First-person video has dominated embodied AI research. Adapting NVIDIA's Cosmos3-Nano world model, DreamVu's new study shows that a single wide-view camera watching the whole scene matched or beat training on all data combined — using half the footage.

PHILADELPHIA, PA — July 2, 2026 — DreamVu today released RetailSMV (Retail Synchronized Multi-View), the first retail video dataset to capture real store staff at work from two synchronized perspectives at once: a head-mounted camera showing what the worker sees, and DreamVu's Alia360 camera observing the entire scene. Alongside the dataset, the company published research demonstrating that the wide scene view — not the first-person view the field has long favored — did nearly all the work when adapting a world model to a real retail environment.

For years, embodied AI research has centered on egocentric data: first-person footage from cameras mounted on a person's head or a robot's body. DreamVu's results invert that hierarchy. Fine-tuning on exocentric footage alone matched or beat training on the full combined dataset on six of seven metrics — while using half the clips.

"The field standardized on first-person data because it standardized on manipulation tasks — and for a hand grasping an object, that's the right view. But a robot working a store doesn't just need to know what its hands are doing. It needs a model of how the whole environment behaves. That signal lives in the wide view, and our results show it's been dramatically undervalued. It's also exactly the view Alia360 was engineered to capture: the entire scene, in 3D, from a single camera."
— Rajat Aggarwal, Co-Founder and CTO of DreamVu

The dataset

Unlike prior retail datasets built around the shopper's experience, RetailSMV captures the staff side of retail: stocking shelves, weighing produce, carrying crates, pushing supply carts, and scanning at checkout — the tasks embodied robots will actually be asked to perform.

Measured on a leading world model

Using LoRA — a lightweight fine-tuning technique that adapts a large model without retraining it from scratch — the team adapted NVIDIA's Cosmos3-Nano world model on RetailSMV. Validation loss fell 2.8×. Generated video improved on every one of 200 held-out test clips, with the statistical gap between generated and real footage shrinking by up to 33.5%.

Paper

The full paper is available on arXiv.

About DreamVu

DreamVu designs and builds patented 360° 3D cameras for robots, and the software that runs on them. The company is headquartered in Philadelphia, PA, with offices in Palo Alto, CA and R&D in Hyderabad, India.