In case of an emergency in a building, first responders need to know current blockages in the building (e.g., blocked passageways and exits) and safe evacuating paths so that the occupants can be guided to the unblocked exits and safe paths toward those exits. To automatically determine blockage levels at buildings, a system that fuses data from multiple sensors and video camera was proposed. A prototype was developed and tested on an experimental model of a pilot building's hallway. A series of damage tests were conducted on the hallway model and recorded by the sensors and the video camera. Individual performances of sensors and video camera were evaluated, and a decision tree method was used to fuse sensor and video camera data for estimating the level of blockage in the hallway for different damage combinations applied on building components. The results demonstrated the technical feasibility of the proposed system and the findings of the decision tree highlight that by using less number of sensors, a cost-effective configuration can be achieved. The estimated blockage information can be used to create a topological map of the damaged building, indicating safe paths toward available unblocked exits. (C) 2014 Elsevier B.V. All rights reserved.