FLIR

When FLIR announced its third-generation FLIR ONE and FLIR ONE Pro thermal camera accessories at CES 2017, the development team faced compounding manufacturing risks. The redesign introduced significant complexity — a patent-pending OneFit™ adjustable connector, a 1.8-meter drop test rating, and VividIR™ thermal imaging — all requiring precise assembly tolerances. Compounding this, FLIR was partnering with a factory that had limited experience in final-product assembly. In electronics NPI, misalignment between design intent and factory execution during DVT can cascade into costly rework, delayed mass production, and field quality escapes. Traditional oversight required engineers to travel to the factory repeatedly, a slow and expensive approach incompatible with a fast-moving product launch.

FLIR deployed Instrumental's computer vision inspection platform directly on the DVT assembly line, with multiple stations configured to capture images of critical assembly states at each process step. Setup was completed in two days — an unusually fast deployment for a new factory environment — and images were available for remote review within minutes of go-live. The system applied computer vision to automatically flag anomalies including improperly fastened screws, misaligned PCB components, unsnapped connectors, and part contamination. Rather than replacing engineer judgment, the platform gave FLIR's cross-functional teams in Santa Barbara simultaneous remote access to real-time assembly data. Multiple conference rooms of engineers — mechanical, quality, and failure analysis — could review the same image record concurrently. FLIR retained the Instrumental stations through mass production to monitor every unit built.

FLIR achieved a 75% reduction in factory travel for the development team while simultaneously accelerating the transition to mass production. Marcel Tremblay, Director of Mechanical Engineering, confirmed the system surfaced issues that would otherwise have gone undetected: "Some of the issues we may never have found." Key outcomes included:

• 75% reduction in factory site visits during NPI
• 2-day station setup time on a new factory line
• PCB defects, mis-assembled components, and fastener failures identified and corrected before production
• Four cross-functional teams leveraged shared image data simultaneously during development
• Instrumental stations retained into mass production for ongoing unit-level quality monitoring

• Remote visual data access can replace most factory travel during NPI — 75% reduction is achievable without sacrificing oversight quality.
• Two-day deployment is realistic for computer vision inspection if the platform is purpose-built for electronics assembly lines.
• Working with an inexperienced factory partner increases NPI risk — automated inspection provides a safety net when tribal knowledge is absent.
• Unit-level image records accelerate failure analysis by giving distributed teams a shared, unambiguous data source.
• Plan for continuity into mass production — the same system that catches DVT issues can monitor yield and quality at full volume.