Design and Development of a Vision Based Human Following Robot with Rocker Bogie Mobility

Abstract

This paper presents the design and development of a low cost autonomous human following robot for indoor service applications. Unlike conventional systems that rely on markers or computationally intensive deep learning models, the proposed approach employs a markerless, vision based tracking framework using Mediapipe for real time human detection and localization. The robot is equipped with a six wheel rocker bogie suspension mechanism to ensure enhanced stability and mobility over uneven indoor surfaces. A lightweight control strategy is implemented in which steering and navigation commands are generated directly from spatial errors in the visual field, enabling effective operation. The mechanical design is validated through finite element analysis and multibody dynamic simulations, while the overall system performance is evaluated using a physical prototype. Experimental results demonstrate stable motion, reliable human tracking, and effective real time response, highlighting the suitability of the proposed system for practical service robotics applications.