Real-time foot attitude estimation for a humanoid robot based on inertial sensors and force sensor

This paper proposes a real-time foot attitude estimation approach consists of a foot attitude acquisition and an auto-calibration method. First, overview of the foot attitude estimation system is presented, including a tri-axis MEMS accelerometer, three angular rate gyroscopes and a data acquisition and processing unit. Then, the arithmetic of data processing and attitude acquisition is introduced. FIR digital filter is proposed to reduce the influence of inherent noise of the sensors, and combination of acceleration and angular rate is adopted to calculate the foot attitude. Finally, a special auto-calibration method based on the force sensor equipped at the ankle of the humanoid robot is presented. The effectiveness of the foot attitude estimation system and the auto-calibration method is confirmed by experiments with our developed humanoid robot.