Illuminating the Black Box: Strategies for Opaque Embedded Systems

Modern elevator systems are no longer simply mechanical machines moving passengers between floors. They have evolved into highly distributed embedded platforms integrating real-time software, motor drives, sensor networks, mobile connectivity, cloud services, and safety-critical control systems. Yet despite this evolution, much of the industry still troubleshoots failures the same way it did years ago—through fault codes, intermittent symptoms, limited logs, and reactive field investigations after the passenger experience has already been impacted. For today’s quality engineering organizations, traditional “Pass/Fail” validation is no longer enough.

In this presentation, I’ll explore how observability-driven quality engineering is reshaping the way embedded elevator systems are tested, monitored, diagnosed, and maintained. Drawing from more than three decades in quality engineering leadership—and over ten years specifically within the elevator industry—I’ll discuss how high-fidelity instrumentation, telemetry mining, synchronized event tracing, edge intelligence, and AI-driven analytics are helping organizations expose hidden system behaviors that conventional diagnostics often miss.

Using real-world case studies—including fluctuating doors, “Phantom Jerks,” and synchronization failures affecting field technicians operating in signal-dead mountain regions—we’ll examine how modern observability strategies can identify degraded conditions long before traditional faults occur. More importantly, we’ll discuss how historical telemetry, instrumentation, warning analysis, and operational data can be transformed into “Golden Profiles” that allow organizations to move from reactive troubleshooting toward predictive operational intelligence.

The presentation concludes with a modern approach to embedded systems quality coverage built around three expanding dimensions:

• Code Coverage
• Environmental Coverage
• Observation Coverage

Because in today’s connected embedded systems, coverage is no longer binary.  The question is no longer simply whether the elevator passed a test.

The real question is:

“How much of the system’s operational health can we actually observe, understand, and prove before the passenger ever notices a problem?”