Touch + vision fusion
Combines tactile streams with vision for true contact awareness.

Tactile-and-vision contact world model that predicts grasp outcomes.
A multimodal world model that fuses touch and vision to predict contact outcomes — hold, slip, crush — a few steps ahead, enabling closed-loop, contact-reactive control.
Built to be GPU-essential, NVIDIA-aligned, and production-ready.
Combines tactile streams with vision for true contact awareness.
Forecasts slip and force a few steps ahead of the grasp.
Feeds Dextra's policy so the robot reacts mid-grasp.
Ingests common tactile sensors through one interface.
Bootstrapped on synthetic contact-rich data from Simdex.
Cuts dropped and crushed picks versus vision-only baselines.
| Model type | Multimodal contact world model |
|---|---|
| Inputs | Tactile + vision streams |
| Edge | Jetson Thor / IGX + Holoscan |
| Training | NVIDIA DGX / HGX |
| Output | Contact outcome + force prediction |
| Integration | Feeds Dextra closed loop |
Start free. Scale to a fleet. Pricing shown is indicative, pre-launch.
Cross-embodiment VLA manipulation foundation model + fine-tuning toolkit.
Explore Dextra →
On-robot edge runtime for low-latency closed-loop manipulation.
Explore Reflex →
Domain-randomized manipulation simulation + data-generation engine.
Explore Simdex →The manipulation layer teams ship with, not a research demo.
One policy runs across many grippers and hands — no per-robot rewrite.
Tactyl predicts slip and force so grasps react before they fail.
Reflex executes on Jetson at control-loop latency, with safety limits.
Every deployment feeds the shared model through the data flywheel.
Benchmarks, sim-to-real notes, and product updates. No hype, no spam.
Tell us your robot and task — we'll scope a pilot.