Foot-Proximal Sensing for Safe Mobility: A Privacy-Preserving Smart Shoe with Calibrated Ultrasonic Ranging and Gait-Aligned Feedback

Independent mobility for individuals with visual impairments is often hindered by ground-level hazards such as curbs, steps, and low obstacles that are challenging to detect reliably with only camera or mid-torso sensors. We present a foot-mounted assistive system that centres perception at the shoe and keeps all navigation decisions on-device to minimise latency and protect privacy. The prototype integrates a toe-box ultrasonic transducer, a Raspberry Pi 3 B+ controller, and vibrotactile/audio cues synchronised to gait; an optional Android viewer provides event-only telemetry under explicit user consent. Methodologically, we make the timing convention explicit and calibrate the ultrasonic pipeline to round-trip time-of-flight with temperature compensation. Bench characterisation over 0.05–0.80 m shows near-perfect linearity (R² ≈ 0.99998) and an implied acoustic speed of ~340 m s⁻¹, confirming end-to-end timing fidelity. In formative indoor walking trials, end-to-end warning latency remained below a 150–200 ms budget while a three-state policy (clear/caution/stop) avoided threshold chatter and maintained cue discriminability. The results indicate that foot-proximal ultrasound, when correctly calibrated and paired with gait-aware feedback, provides fast, illumination-agnostic awareness of trip hazards without imposing continuous video streaming or cloud reliance. We conclude with a discussion of limitations (pilot scale, mostly indoor scenes) and outline a path to a larger, protocolized study that includes outdoor conditions, standardised usability metrics, and optional sensor fusion.