Vision-Guided Robot Integration Checklist (ISO 13849-1 & ISO/TS 15066 Compliant)

Vision-guided robot integration combines machine vision (e.g., 2D/3D cameras, lighting, calibration, feature detection) with industrial robot control to enable adaptive tasks such as bin-picking, part alignment, or dynamic path correction. However, introducing vision into the safety-related control system (SRCS) significantly increases complexity: latency, image misclassification, occlusion, calibration drift, and sensor failure can all compromise safe motion execution. ISO 13849-1 mandates a systematic approach to achieving required Performance Levels (PLr) via architecture category, MTTFd, DC, and CCF analysis—applied here not only to traditional safety components (e.g., light curtains) but also to vision subsystems acting in safety functions (e.g., 'vision-based presence detection' or 'vision-validated stop'). ISO/TS 15066 further constrains integration when humans share the workspace, requiring validation of transient contact forces, pressure thresholds, and time-weighted energy limits—especially critical when vision enables proximity-aware motion (e.g., hand-guided teaching or speed-scaled tracking). Therefore, the checklist ensures traceability from hazard identification (via risk assessment per ISO 12100) through safety requirement specification, architecture design, validation testing (including worst-case illumination, motion blur, and adversarial object conditions), and documentation (e.g., safety validation report, SIL/PL calculation evidence, and vision subsystem validation records). It also mandates verification of synchronization between vision trigger timing, image acquisition latency, pose estimation uncertainty, and robot controller response—since end-to-end delay directly impacts stopping distance and force compliance under ISO/TS 15066 Annex A.