A robust humanoid robot navigation algorithm with ZUPT

Abstract

This paper discusses algorithmic concepts, design and testing of a pedestrian dead reckoning (PDR) navigation system based on a low-cost inertial measurement unit (IMU) attached to a user's shoe. The algorithm uses the technique known as “Zero Velocity Update” (ZUPT) and Kalman Filter consists of 24 error states to reduce IMU errors. We propose a novel dynamic and more robust algorithm to detect the stance phases during walking. The system works well in both 2D (2-dimensional) and 3D environments. Test results show that its horizontal positioning errors are always below 0.3% of the total travelled distance, and the vertical errors are below 0.7%, even on 3D terrain. These results reach the highest position accuracy in available literature.