Enhancing Agricultural Productivity: A Sensor Fusion, Spectroscopic Near Infrared, and AI-Based Framework for Monitoring Fruit Ripeness

Fruit ripeness evaluation is essential for harvest optimization and post-harvest loss prevention. Traditional visual inspection is inefficient. This study employs sensor fusion and AI algorithms to evaluate fruit maturity in real-time using gas sensors (i.e., MQ-3, MQ-6, MQ-8, MQ-135) and Near-Infrared spectral sensors (i.e., GY-AS7263). Primarily, data pro-cessing-sensor fusion, data preparation, and AI model application-is crucial to the proposed framework. Subsequently, sensor fusion approaches improve data accuracy by combining signals from various sensors. Secondly, data pretreatment techniques including noise reduction, normalization, and feature extraction prepare data for analysis. Thirdly, we use powerful machine learning algorithms such as Random Forest to judge maturity more accurately and reliably than conventional techniques. Lastly, integrating NIR spectroscopy with gas sensor data resolves cost and model transferability issues, enhancing accuracy and acces-sibility for agricultural use. Static image processing techniques provide challenges for real-time applications, including expensive computer costs and lower reliability. The proposed framework based on multi-sensor fusion system enhances agricultural applications' versatility, scalability, and reactivity, surpassing existing methodologies.