CVPR 2026 • 1st SOS Workshop • Full-day Workshop • June 3rd • Denver, Colorado, USA
Sense of Space: Multi-Sensory Modeling for Embodied Intelligence
in conjunction with CVPR 2026
June 3rd
Date
8:30 AM – 5:00 PM
Time
Mile High 2C
Room
Hall A #275–282
Posters
with the first workshop focusing on emergent special sensory modalities, and dexterous hand-object manipulation.
Workshop Recording
Workshop overview
Embodied intelligence in humans and robots relies on the integration of multiple sensory modalities — vision, touch, sound, proprioception, bio-signals, and beyond — into coherent spatial understanding and actionable policies. While recent advances in AI have significantly improved language and visual perception, progress in real-world embodied intelligence remains fundamentally constrained by our ability to model, align, and learn from multi-sensory physical interactions.
Sense of Space is a full-day CVPR 2026 workshop that brings together researchers working across sensing hardware, data acquisition, representation learning, and embodied applications. To provide focus within this broad and critical space, the 2026 edition centers on emergent special sensory modalities, and dexterous hand-object manipulation.
The workshop emphasizes challenges including novel sensing modalities, data-driven sensor design, large scale synchronized multi-sensory data collection, cross-modal spatial–temporal alignment, robustness to missing or noisy modalities, and generalizable representations for real-world manipulation in robotics & XR.
Resource: Keynote, Oral and Panel Discussion
Recording timestamps (H:MM:SS)
| Timestamp | Session | Video |
|---|---|---|
| 0:10:35 |
Nima Fazeli
Tactile Intelligence
|
Watch |
| 0:46:23 |
Mahi Shafiullah
Robots are slowing down robotics
|
Watch |
| 1:05:20 | Oral Session — Smell and Touch | Watch |
| 2:02:08 |
Marc Pollefeys
Hand, Object, Touch and Force
|
Watch |
| 2:39:10 |
Wojciech Matusik
Tactile Sensing for Human Machine Interaction, Computational Design and Fabrication
|
Watch |
| 3:22:00 |
Homanga Bharadhwaj
Observational Learning for Manipulation via Visual Imitation of Humans
|
Watch |
| 3:53:19 |
Boyi Li
Learning Multimodal Robot Policies with Generative Signals
Video omitted at speaker's request.
|
Watch |
| 3:55:15 |
Sha Yi
When is touch necessary for robotics?
Audio unavailable in the video, see slides link here.
|
Watch |
| 4:28:15 | Oral Session 2 — Special Sensors | Watch |
| 4:58:41 |
Lingjie Liu
Seeing Space Before Touching It
|
Watch |
| 5:33:51 |
Paul Liang
Expanding AI's sense
|
Watch |
| 6:06:28 |
Panel Discussion
Rao Fu, Sha Yi, Homanga Bharadhwaj, Christian Theobalt, Mahi Shafiullah, Ismini Lourentzou, Paul Liang
|
Watch |
Keynote Speakers / Panelists
Speakers spanning sensor design, systems, algorithms, and applications.
Organizing Committee
Rao Fu
Brown University
Li Guan
Meta Reality Labs
Kaichen Zhou
Harvard / MIT Media Lab
Kun He
Meta Reality Labs
Tomas Hodan
Meta Reality Labs
Ergys Ristani
Meta Reality Labs
Eric Yi
Tsinghua University
Jessica Yin
NVIDIA
Qi Ye
Zhejiang University
Yichen Li
MIT CSAIL
Yining Hong
Stanford University
Devin Murphy
University of Washington
Ray Song
MIT Media Lab
Xueyan Zou
Tsinghua University
Yunzhu Li
Columbia University
Hao-Shu Fang
University of Maryland
Ruoshi Liu
University of Maryland
Yiyue Luo
University of Washington
Vatsal Mehta
Meta Reality Labs
Mengyu Wang
Harvard University
Wojciech Matusik
MIT CSAIL
Paper Committee
Thanks to all of our paper committee members.
Kun He
Kaichen Zhou
Ziyi Kou
Xueyan Zou
Tomas Hodan
Yichen Li
Eric Whitmire
Devin Murphy
Yuxin Ray Song
Yiyue Luo
Yue Liu
Ergys Ristani
Li Guan
Rao Fu
Jessica Yin
Ankit Kumar
Qi Ye
Dingtian Zhang
Hao-Shu Fang
Yunzhu Li
Ruoshi Liu
Sponsors
Sponsorship opportunities
Support poster awards, student travel, live streaming, and events — and connect with a concentrated
audience of embodied AI researchers and practitioners. Interested in supporting? Reach out.