A user-specific activity pattern generation framework for evidence-based ABW planning

To plan successful activity-based workplaces (ABW), architects need to clearly understand user-specific activity patterns through the accurate recognition of user activity. Because user activity is closely associated with space, equipment, and users, such diverse activity-related information should be essentially considered for accurate activity recognition. However, previous activity recognition methods have limitations for accurately recognizing user activity for ABW planning, because they only relied on sensor-estimated data and are, therefore, unable to comprehensively consider diverse activity-related information. The study thus integrates site investigation and sensor estimation using a Bluetooth Low Energy beacon and accelerometer for accurately recognizing user activity based on diverse activity-related information. We defined five important items of activity-related information (user actions, number of nearby users, function of space, equipment located in space, and space use policy) and developed a user-specific activity pattern generation (UAPG) framework consisting of three stages: (1) the preparation stage, (2) sensor-estimation stage, and (3) activity pattern generation stage. The demonstration was conducted through scenario-based experiments in an academic office building. In the demonstration, the UAPG framework achieved 91.8% of activity recognition accuracy and successfully generated user-specific activity patterns. In addition, information regarding space usage, proportion of activities, and spatial preference of the user was generated based on a user-specific activity pattern. Such objective information provided by the UAPG framework enables evidence-based ABW planning that efficiently accommodates users with minimal office space, while simultaneously increasing their satisfaction and productivity.