Velocity Estimation for Dog's Moving Trajectory using its Gait and IMU

Abstract

When searching victims in disaster areas, information of trajectory of search and rescue dogs is needed to access the victims. We developed a method to record and visualize the dog's trajectory that incorporates an inertial sensor to estimate the dog's velocity and position, even in environments where GPS signals are not available or accurate, such as collapsed buildings, mountain areas, etc. The dog's velocity and position are estimated by integrating accerelation signals measured with the inertial sensor. An accumulative error of the velocity estimation is corrected just when lateral or vertical velocity is zero. The timing of these zero velocity timings are estimated by using the inertial sensor data. Estimated velocity and position by the proposed method is more accurate than those estimated by the other methods without correction of velocity or position.