Autonomous Active Sensing in Earthquake Victim Search

Objectif

Rapid detection of survivors after an earthquake is critical to reducing fatalities, yet today’s search teams still rely on manual sensor placement, static probing, and intuition-based coverage. While sensor technologies such as microwave, seismic or acoustic, and visual systems exist, there is no established method for how to optimally place, move, and fuse them during real search operations. The AASE project will develop a new approach to earthquake victim search, using information-based active sensing strategies to guide first manual and later autonomous searches, taking into account region-specific conditions as well as universal principles applicable to earthquake response worldwide. Active sensing with cameras and seismic sensors specific to search missions will collect high-value data, which will then be fused in real time to inform rescuers about the likely positions of survivors, even in uncertain environments. From the outset, AASE will incorporate field knowledge from both governmental and internationally recognized non-governmental organizations, through qualitative studies and joint testing. Theoretical models and simulation-based validation will take place at the Middle East Technical University (METU) Center for Robotics and Artificial Intelligence (ROMER), followed by the Ministry of Interior’s physical disaster training grounds. In the final phase, ROMER’s legged robots will assist search operations, augmenting human capacity and reducing safety risks in the rubble. The fellow brings expertise in sensing and dynamic systems from the University of Washington and the University of Nevada, Reno, and through AASE will gain project management experience and emerge as an interdisciplinary academic in sensor-driven applications across Europe.