Double Lower Limb Kinematics-based Invariant Extended Kalman Filtering for Pedestrian Inertial State Estimation

This paper combined the double lower limbs and waist as rigid bodies. Each angle of hip, knee, and ankle joint can be measured through several Inertial Measurement Unit (IMU), formulating the double lower kinematics. Focusing on the model uncertainty of stance and swing phase switching within gait cycle, a novel invariant extended Kalman filter (EKF) was contrived to produce state estimation. Towards the rotation path, the people executed walking locomotion in simulation. Compared to the classical EKF, the more rapid error convergence for pedestrian state estimation is accomplished using the invariant extended Kalman filter (InEKF). This methodology demonstrated the validity, feasibility and accuracy of pedestrian inertial state estimation with InEKF, which expanded the application scope of inertial measurement apparatus.