State estimation of mobile robot using GPS signal that includes reflected and diffracted waves

We propose a method to improve the state estimation accuracy of mobile robots placed near high-rise buildings using the statistical property of the reflected and diffracted waves of global positioning system (GPS) signals. First, it is assumed that a GPS signal that contains a reflected and diffracted wave is denoted by the sum of the true position information and noise that follows a time-varying Gaussian distribution. On the basis of this assumption, the time-varying bias of a GPS signal is tracked using a Kalman filter. In addition, a particle filter, which executes sampling and likelihood evaluation using the estimated bias, is developed. With the proposed method, a GPS signal that contains the rejected noise introduced by the conventional method can be used efficiently, and the state estimation accuracy of the robot in a shadow area of GPS satellite can be improved. Furthermore, a control system for an autonomous mobile robot incorporating the proposed state estimation mechanism is developed, and its effectiveness is evaluated via simulation.