Multisensor Data Fusion for Determining Hallway Blockages in a Building during Evacuation

When a building is under the threat of a disaster/an emergency, it is important to rapidly retrieve up-to-date indoor navigation conditions by identifying the hallways that are blocked. This paper investigates the contribution of data from each sensor that is deployed inside a building in accurately determining the current blockage conditions at hallways. Sensor performances were assessed both individually and by fusing multisensor data at the building element-level. The final goal is to create and visualize evacuation and navigation paths for the first responders and the occupants during a disaster/an emergency. A prototype was developed to integrate Building Information Model (BIM) with hallway blockage information retrieved from multiple sensors and to provide guidance in a building based on the building’s current status using graph theory principles. This paper specifically discusses if less sensors could be used to get similar accuracy results in detecting hallway blockages based on the individual and fused data from multiple sensors. INTRODUCTION Obtaining timely and accurate information related to a damaged building is critical when a building is under the threat of a disaster/an emergency. This information will be used to guide building occupants safely and quickly out of a building. Previous experience shows that even panicked occupants of a damaged building tend to comply with the authorized visual and/or audible guidance, as observed in the case of 9/11 terrorist attack in which the victims followed the directions of 911 operators’ (National Commission 2004). Also, the emergency