NVIDIA at CVPR 2026

4–8 p.m. MT

The Alpamayo Summit will explore how end-to-end reasoning models, open datasets, simulation, and safety frameworks are accelerating scalable, trustworthy Level 4 autonomy and shaping what the autonomous driving ecosystem builds next.

9:15 a.m.–6 p.m. MT

This workshop will explore how virtual humans and robots can learn from one another to advance embodied agents, with Umar Iqbal presenting a keynote on human digital twins, robot skill learning, and cross-disciplinary challenges such as retargeting, contact modeling, and foundation models.

1–6 p.m. MT

This workshop will explore graphics-in-the-loop, physically grounded 4D reconstruction for physical AI digital twins, with Francis Williams presenting a keynote on bridging graphics, vision, and robotics for real-to-sim-to-real physical AI.

10 a.m.–6 p.m. MT

NVIDIA Booth 211 opens with short technical demos and theater talks across NVIDIA Cosmos, robotics, visual inspection, autonomous driving, and partner-led AI workflows.

1–2:15 p.m. MT

PPISP explores physically plausible ways to compensate for photometric variation in radiance fields, improving how dynamic lighting and appearance changes are handled in 3D reconstruction.

6–10 p.m. MT

NVIDIA’s Researcher Celebration brings CVPR researchers together for an evening of food, drinks, networking, and community building with fellow AI innovators.

10 a.m.–6 p.m. MT

Saturday’s booth theater focuses on NVIDIA Alpamayo, worldbuilding, AI models, and robotics demos, with technical sessions from NVIDIA researchers and ecosystem partners.

12–1:30 p.m. MT

Launching Legends connects researcher-founders with VCs, NVIDIA Inception, mentors, and cloud partner leaders to help turn breakthrough research ideas into startup pathways.

2–3:15 p.m. MT

NitroGen introduces an open foundation model for generalist gaming agents, trained across large-scale gameplay data to transfer skills across diverse game environments.

10 a.m.–6 p.m. MT

The final expo day features a shorter demo and theater program focused on CWE and autonomous-driving partners before the show floor closes.

3:30 p.m.–05:30p.m. MT

This poster presents a latent reasoning approach for autonomous driving, using world-model and action tokens to improve end-to-end planning and closed-loop performance.