Development and Evaluation of a Smart Vacuum Robot with Adaptive Navigation

Abstract

Autonomous vacuum cleaners promise hands-free hygiene, yet most commercial units are engineered for spacious Western interiors and are ill-suited to the narrow corridors and mixed floor materials typical of Vietnamese homes. A low-profile, 350 mm‑diameter robot was designed and fabricated, featuring a dual-stage centrifugal vacuum system, differential‑drive kinematics, and a Robot Operating System 2 (ROS 2) autonomy stack built around an RPLIDAR A1M8 sensor and Hector Simultaneous Localization and Mapping (SLAM). Quantitative trials, conducted in a 90 m² apartment testbed, assessed cleaning efficiency, battery endurance, and navigation accuracy. The prototype achieved mean cleaning efficiencies of 97.4 ± 1.9% on smooth tile and 92.0 ± 3.4% on 10 mm‑pile carpet, statistically surpassing the benchmark (p < 0.05). Battery runtime averaged 72 ± 3 min (quiet), 64 ± 2 min (standard), and 48 ± 2 min (max), whereas mean localization error remained below 2.1 ± 0.7 cm. The results verify that a cost-effective sensor suite, coupled with energy-aware path planning, can deliver competitive performance in the cluttered layouts of Vietnamese dwellings, laying the groundwork for a domestic robotic appliance supply chain.