PoseSentinel

PoseSentinel is a desktop + web AI app that continuously monitors a robot’s camera pose and vision health in production without requiring new calibration routines. It ingests live frames plus robot kinematics (or fiducial snapshots when available) and estimates whether the camera-to-robot transform is drifting, vibrating loose, or slowly shifting due to temperature and impacts. Instead of vague “vision accuracy dropped” alerts, it outputs actionable diagnostics: likely axis of shift, confidence, when it started, and which station/robot is affected. It also generates a short incident report with frame evidence and recommended checks (mount torque, bracket fatigue, lens focus, lighting changes). This is not a research toy: it focuses on pragmatic thresholds, simple integrations (ROS/OPC UA), and clear ROI by preventing scrap, downtime, and firefighting. Expect some false positives early; the value is catching real drift days earlier than humans.