Sensor Fusion Method for In-Cabin Humidity Prediction

Maintaining optimal in-cabin humidity levels is part of occupant comfort, air quality, and the effective operation of climate control systems, particularly for functions like windshield defogging. This paper introduces a novel sensor fusion methodology for predicting in-cabin humidity without the need for a dedicated humidity sensor. The proposed approach leverages readily available vehicle data, integrating information from ambient temperature sensors, in-cabin temperature sensors, occupant detection systems, window status (open/closed), and climate control settings. By intelligently fusing these diverse data streams, a predictive model is developed to infer the dynamic humidity conditions within the vehicle cabin. We discuss the complex interactions between these parameters, such as the moisture contribution from occupants, the influence of external air ingress through open windows, and the dehumidifying or humidifying effects of the HVAC system. The paper details the development and validation of the predictive algorithm, highlighting its capability to estimate humidity levels under various operational scenarios. Challenges in modeling the transient and non-linear relationships between inputs and humidity, as well as the evaluation of the model's accuracy against ground truth data, will be presented. Initial results demonstrate the feasibility and robustness of this sensor fusion approach, offering integrated solution for intelligent cabin climate management and enhanced occupant experience.