Hardware Limitations

Phase 12 records do not change the hardware status: zero physical screenshots, photos, diagnostics, or measurements were supplied. Every named device workflow remains Blocked or Not tested. Consult the result pages under Physical Validation before interpreting any capability.

Phase 11 does not convert simulator evidence into hardware evidence. iPhone 13 Pro camera/sensors/power/thermal behavior, iPad Air M1 radio/touch/accessibility behavior, GoPro HERO10 connection/GPMF variants, and DJI Mini 4 Pro codecs/sidecars remain Not tested until their physical plans are recorded.

Phase 2 implements the rear iPhone wide camera through AVFoundation, preferring 1080p at 30 FPS and gracefully accepting the closest supported session preset/format. It has not been validated on every iPhone model. Microphone input defaults off. GoPro Hero 10 and DJI Mini 4 Pro remain explicitly unsupported.

Simulator sessions contain generated video and deterministic sensor values and are labeled simulatedCamera; they do not validate lenses, stabilization, permissions, GPS, IMU accuracy, interruptions, thermal behavior, storage throughput, or battery life. No claims are made for vendor protocols, Core ML inference, live transfer, or frame-perfect sensor synchronization. Phase 3 simulator connections are labeled and use an in-process lab because multi-simulator Bonjour can be unreliable. Physical Bonjour discovery, local-network permission, Wi-Fi isolation, firewall behavior, throughput, latency, radio loss, background behavior, and VideoToolbox hardware encode/decode require two provisioned devices. HEVC is preferred by negotiation and H.264 is the fallback; simulator preview frames are deterministic monitoring data rather than camera pixels.

Continuous video and 20 Hz motion sampling consume battery, storage, and processing capacity and may heat the device. Use a ventilated, secure dashboard mount that does not obstruct vision, controls, airbags, or legally required visibility. Connect power only with a safe routed cable. Configure and start recording while parked; never operate the application manually while driving.

The iPad console must be operated only by a passenger or while parked. Connection loss, latency, preview freezes, and stale telemetry are expected failure modes and must never be treated as a safety signal or affect vehicle operation.

Phase 4 validation limits

Catalog, integrity, import-directory, export, and chunked-transfer behavior can be tested with temporary files and simulator records. Physical iPhone-to-iPad transfer throughput, background continuation, Files/share-sheet permissions, device storage pressure, bookmark persistence, and drag-and-drop from Finder require real devices or interactive platform testing and are not claimed validated.

Phase 5 validation limits

Deterministic detection, tracking, scoring, storage, indexing, and overlay behavior is testable without hardware. No trained third-party model is bundled. Physical pixel-buffer inference, real Core ML accuracy/compute behavior, thermal adaptation, battery impact, orientation metadata, AVAsset offline decoding, and production annotated-video rendering remain unvalidated. See Phase5Limitations.md.