Machine Learning-Based Sensor Synchronization for Enhanced Spatial Awareness in Assistive Technologies

Enhancing spatial awareness for the visually impaired is crucial for autonomy and mobility. Our research addresses this by proposing a novel approach to sensor data synchronization for assistive technologies, re-framing the challenge as a forecasting task suitable for machine learning solutions. We developed a hard hat assembly equipped with 2D LiDAR, dual 1D LiDARs, and a depth camera to create comprehensive spatial awareness for navigation. By applying various machine learning algorithms to predict 2D LiDAR readings from higher frequency 1D LiDAR data, we aimed to overcome conventional synchronization limitations. Our investigation revealed that while linear models and some non-linear approaches were inadequate, the Generalized Auto-regressive Conditional Heteroskedasticity (GARCH) model showed promise in capturing short-term volatility. Our findings contribute to assistive technologies by introducing a machine learning paradigm for sensor synchronization, potentially improving spatial awareness systems’ accuracy and reliability. This system has the potential to significantly enhance navigation for the visually impaired, fostering greater independence and safety.