Virtual Lab
Virtual Lab Mobile / Sensor Node
Edge acquisition system and mobile scientific instrumentation codebase investigating how commodity sensing hardware can participate in structured, attributable scientific acquisition.
Purpose
Transform mobile hardware (cameras, IMUs, RF radios) into calibrated scientific acquisition instruments that strictly preserve the epistemological distinction between raw observation, calibrated measurement, and algorithmic inference.
How it works
Built in Kotlin targeting Android 14 (API 34). Operates as an edge acquisition node acquiring multi-sensor streams under an Android Foreground Service (AcquisitionForegroundService). Integrates Camera2 API with remote focus/exposure controls over WebSockets, sub-meter Wi-Fi RTT (802.11mc) ranging, and 100 Hz calibrated IMU/magnetometer sampling, enforcing the universal threefold classification: ● OBSERVED (direct raw pixel/sensor reading), ◆ MEASURED (calibrated calculation), and ▲ INFERRED (model-derived prediction).
- The Epistemological Triad: Strictly separate observed physical readings from calibrated measurements and model inferences.
- Deterministic Sampling: Monotonic clock-synchronized sampling across camera frames, IMU bursts, and Wi-Fi RTT sweeps.
- Continuous Edge Acquisition: Uninterrupted sensor streams maintained via Android foreground services with persistent notifications.
- Remote Instrument Control: Real-time telemetry and camera adjustments commanded bidirectionally over WebSockets.
Key findings
Triad Integrity Value
SUPPORTEDRigorous separation of Observed, Measured, and Inferred fields is necessary to identify when models hallucinate over degraded inputs.