Concept / bus

On-Device Inference

Running a model on the robot or board itself (NPU or GPU), not in the cloud — lower latency, works offline, no per-seat bill.

Connections

3 relationships

How this entity connects across the Physical AI graph.