The stack underneath a robot that works
A robot is a piece of hardware. What makes it productive is the layer below it: navigation, perception, task queues, data and supervision.
We do not build robots. We build the connection.
Manufacturers solve walking, grasping and vision. What stays open is the connection to the real world of a factory: which task, in which order, for which batch, approved by whom and logged where.
That is where we work. We use ROS 2 where it is available, manufacturer SDKs where it is not, and build a thin orchestration layer on top that talks to your systems.
Four layers, one system
Hardware and sensors
Robots, grippers, cameras, LiDAR, force and environmental sensors.
Control and perception
Navigation, obstacle avoidance, object recognition, task execution.
Fleet orchestration
Task queues, priorities, allocation across units, supervision.
Business systems
ERP, WMS, MES, traceability, reports and dashboards.
Five rules we do not break
Your data stays yours
Your infrastructure is the default. Cloud only by explicit agreement.
No vendor lock-in
Everything we configure is documented so someone else could continue.
The human has the last word
Every autonomous action has a stop point and a manual override.
Open standards first
ROS 2, OPC UA, MQTT and REST before closed solutions.
Documentation from day one
The installation is handed over with drawings, settings and instructions.
Autonomy stops where risk begins
Every installation has zones: where full speed is allowed, where it is reduced, where entry is forbidden while the machine moves. That is not software, it is spatial design.
Vision systems record only what the task requires. Where there is a risk of recording faces, we define it explicitly with you and document it in the data protection file.
Want a technical conversation?
We talk to your engineers, not only to purchasing.