Intelligent Ergonomic Chair with AI-Driven Posture Analysis and Correction with Real-Time Feedback

Abstract

Prolonged sedentary behavior in modern workplaces is a primary catalyst for musculoskeletal disorders (MSDs). Existing ergonomic chairs provide only passive support, while wearable posture trackers are intrusive and non-scalable for daily use. This paper presents an intelligent ergonomic chair: a noninvasive, active Internet of Things (IoT) system that uniquely integrates embedded pressure sensing, deep learning classification, closed-loop pneumatic actuation, and generative AI advisory into a single unified platform. The system utilizes a 10-channel Force Sensitive Resistor (FSR) array to map seated pressure distribution. A custom ESP32-based architecture processes and transmits telemetry via MQTT to a Multilayer Perceptron (MLP), achieving 99.5% classification accuracy across eight distinct posture classes when evaluated on five held-out participants unseen during training. Upon detecting sustained poor posture, the system autonomously actuates a closed-loop pneumatic lumbar mechanism to restore spinal alignment without physical user intervention. A real-time dashboard further integrates the Google Gemini API to transform session statistics into personalized physiotherapeutic insights. A within-subject usability study (n = 30, 15-minute sessions) demonstrated an 82.0% statistically significant reduction in sustained poor posture (p < 0.001), validating the system's efficacy for proactive ergonomic health management.