FURI | Spring 2026

Scout to Guide: Multi-Agent Assistive Navigation on a Quadruped Robot

Data icon, disabled. Four grey bars arranged like a vertical bar chart.

Over 253 million people worldwide suffer from visual impairments, yet most assistive devices lack customizable navigation capabilities. This research proposes a framework for quadruped robot navigation using multi-sensor fusion of RGB and LiDAR sensors, integrated with object detectors, vision language models, and a Bayesian Optimization library to determine optimal navigation thresholds. This approach aims to enhance human-robot guidance by enabling robots to interpret scenic environments and deliver tailored routing instructions, potentially improving assistive robotic navigation and making human-robot interactions safer.