Dead-Reckoning Method Using the Spring Model

This paper proposes a tracking method of pedestrian trajectory for a portable device that has a 3-axis acceleration sensor, a 3-axis magnetic sensor, and a GPS receiver. The proposed method can achieve higher positioning accuracy compared with conventional pedestrian navigation technique using GPS with low power consumption. Low power consumption was achieved by intermittent activated GPS. The trajectories between the discrete position information acquired with the intermittent activated GPS were interpolated by original pedestrian trajectory estimate method using the acceleration sensor and the magnetic sensor. The acceleration sensor is used for a pedometer to estimate the length of trajectory, and the magnetic sensor is used for a corner detector to know the terminal of the trajectory. Measures of the magnetizing of the portable device (ex. cell-phone), is given to the corner detector. The trajectories of the target person are estimated by a dead-reckoning method by using these sensors. Each trajectory is approximated to straight line (link) using the characteristics that the walker walks in a straight line. The proposed method works to move the estimated link to the most plausible place. But it is hard to move these links to the optimal place due to GPS accuracy. Therefore, the authors use the “Spring Model”, which attracts the position of the link by virtual springs to solve this problem. Three numeric optimization methods were evaluated by the simulation to settle the Spring Model. Simulation results showed that the “Steepest Descent Method” was the best way. Another simulation result showed that the proposed method has a tracking ability of pedestrian trajectory for a portable device, and was also shown reducing the number of GPS activation to less than 5% of conventional pedestrian navigation technique using a GPS receiver.