Occupational safety in confined, metal-dense environments such as water treatment plants (WTPs) requires high-precision monitoring to mitigate risks such as falls and toxic gas exposure. In this study, a sensor-driven monitoring system was designed for narrow, enclosed reverse osmosis water treatment rooms. The system tracks localized environmental parameters to prevent equipment failure and ensure operator safety in confined high-humidity zones. We developed a multi-functional safety helmet integrating ultra-wideband (UWB) for positioning, a nine-axis inertial measurement unit for kinematic analysis, and a gas-sensing network. To overcome the non-line-of-sight and multipath interference typical of industrial geometries, an extended Kalman filter was implemented to fuse UWB ranging with pedestrian dead reckoning. The test results demonstrated a battery endurance of 12.5 h, successfully operating for full work shifts. The fall detection algorithm identified distinct kinematic phases, recording an impact peak of 11.59 g. Field testing in a WTP in Taipei, Taiwan, showed that the fusion algorithm smoothed trajectories compared with erratic UWB trilateration. Environmental sensors maintained real-time monitoring of chlorine (0.07–0.09 ppm) and methane (200–300 ppm) concentrations, both well within safety thresholds. This research advances sensor technology by demonstrating a robust, multimodal architecture that ensures centimeter-level accuracy and reliable anomaly reporting in hazardous narrow-space environments where single-source sensing traditionally fails.