Rhoda AI demonstrates FutureVision DVA model on real-world manipulation workflow
Evidence notes
- Rhoda AI demonstrated its FutureVision Direct Video Action model on a real-world manipulation workflow, showing the system completing the task without human intervention.
- The company framed the model around learning physical behavior from large-scale video pretraining rather than relying only on narrow task-specific robot data.
- The demonstration was presented as evidence that Rhoda's video-based control approach could handle higher-variability robotic work beyond tightly scripted lab settings.
Company context
Rhoda AI builds general-purpose foundation models for robotics, using web-scale video pre-training and robot post-training to drive adaptive robotic behavior in commercial and industrial environments. Its work centers on Direct Video-Action Models, long-context visual memory, one-shot learning, and real-world robot task execution.