Given the global population’s significant growth in the aging demographic, there is an urgent need for advanced technologies capable of monitoring the daily activities of older adults. Active Assisted Living, which integrates multimodal sensors, has emerged as a promising solution to promote independent living among older adults. However, the lack of an efficient fusion method has hindered the full potential of Active Assisted Living technology. This study aims to develop a novel multimodal sensor fusion framework utilizing diverse smart home devices to achieve accurate recognition of daily living activities for older adults. To validate the effectiveness of the framework, a case study was conducted involving 5 individuals in a carefully constructed smart home environment. Five machine learning models were employed to evaluate the recognition performance across 23 indoor activity scenarios. The evaluation results demonstrate the effectiveness of the sensor fusion framework, with an average recognition accuracy exceeding 0.850, and the highest accuracy achieving 0.958.