embedded ocean GmbH has released version 2.1 of their software-defined automation platform Xentara. The update adds image processing to the platform’s core capabilities, allowing users to handle visual input like any other data source. Incoming camera pictures can be fed directly to an ONNX machine learning model running inside the same runtime, and the model’s output can trigger actions in the real-time control core. This not only allows for faster and more reliable optical recognition systems at industrial scale (e.g. for quality control) but is also another step towards Physical AI – machine vision leading directly to action. The update also introduces the new GenICam Skill, allowing users to easily integrate most available industrial camera systems.
Released together with the new version is the Xentara Inspector, a web based graphical interface for inspecting the data model, reading and manipulating values at runtime. Access is role based and secure, so you decide who gets to see what data.
Besides various performance enhancements, the update also adds compatibility for Debian Trixie.