Design and feasibility validation of a mobile home robot system for multimodal health monitoring

Abstract

To address the issue that early signs of cardiovascular and cerebrovascular risks are often overlooked in daily home life, this paper designs a mobile home robot system for multi‑modal health monitoring. The robot adopts a three‑segment mechanical structure consisting of a cone base, a lifting column, and a smart mirror head. It integrates autonomous navigation, vision‑based facial asymmetry detection (for stroke risk indication), non‑contact temperature measurement, touch‑based SpO₂/heart rate monitoring, and medication management. Feasibility is validated through two sets of simulations. First, SolidWorks simulations show that within the lifting height range of 40–75 cm, pitch ±30°, and yaw ±90°, no fatal interference occurs; the overturning safety factor under the most dangerous posture reaches 4.2, and the motor driving torque is sufficient. Second, experiments on the public Goldenhar‑CFID dataset demonstrate that the random forest classifier for facial asymmetry achieves 93.2% accuracy (F1 = 0.9502) on valid detection samples, maintains accuracy above 86% under ±15° pitch variations, and above 87% under low‑light and bright‑light conditions. These results confirm the mechanical reliability and core vision‑based health algorithm feasibility of the proposed robot. Future work will focus on full prototype integration and long‑term home trials.