Skip to content
Use with AI

Vision & AI ​

Build a vision workflow from images, reviewed labels and local models, then use its observations in your application. Choose Vision Studio for inspection tasks or explore VLA models for observation-to-action experiments.

Concept illustration: PLEXA One with symbolic visual observations and a small workpiece.
EXPLORE THE CONCEPTInteractive teaching model
PERCEPTION TASKPosition + orientation + frame

Synthetic scene · rule-based demonstration

Evaluate the synthetic scene to compare the output.

General 6D pose estimation is a perception task; this planar example does not establish an implemented 6D adapter.

Local RGB workflow

Vision Studio

Collect images, review labels, train local models and inspect good, bad or uncertain decisions.

Explore
Scripted application

Use observations in code

Read typed observations, branch on inspection results and prepare proposals before Runtime preview.

Coming soon
Learned policies

Vision-language-action models

Compare model-specific observations and action formats, then prepare a matching simulation experiment.

Explore
Bounded tools

Agents & MCP concept

Understand read, preview and command-tool responsibilities and follow the concrete linked guides.

Coming soon

Quality decisions · Coming soon explain detection versus inspection. FiftyOne · Coming soon explores stored results; CVAT · Coming soon supports reviewed annotation changes. The policy integration checklist · Coming soon keeps decoding, frames, units and execution ownership explicit.