A robust pedestrian navigation algorithm with low cost IMU

Abstract

Zero velocity update (ZUPT) is an effective way for pedestrian navigation in a GPS (Global Positioning System) denied environment. The stance phase in each step provides zero velocity measurement for IMU (Inertial Measurement Unit) drift correction. Most previous research, however, gives navigation solutions only for pedestrian walking but not running. Compared with walking, running has a shorter stance phase with qualified as zero velocity. Therefore a stance phase detector for walking may not be capable for running. This paper presents a novel ZUPT algorithm which can achieve robust pedestrian navigation for walking, stair climbing, and running. Our stance phase detector consists of one footstep detector and two zero velocity detectors (ZVDs).The footstep detector is used to mark each new step, and the first ZVD (ZVD1) can successfully detect zero velocity while walking by setting thresholds on both gyroscope and accelerometer measurements. While ZVD1 is failed for running, the second ZVD (ZVD2) is introduced with a relative larger threshold on gyroscope measurement only. The proposed stance phase detector was tested for walking, running and stair climbing. In all cases, most of the footsteps are detected correctly and our ZUPT algorithm can be successfully implemented. Experimental results show that the navigation accuracy of the proposed algorithm for running cases is comparable to that of walking only cases. Tests on a biped robot are being also conducted to verify the effectiveness of the algorithm.