Parameter identification method for a three-dimensional foot–ground contact model

•The optimization approach allows adjusting the parameters of the contact model.•The model provides good results for normal force and horizontal torque components.•The motion obtained through forward dynamics is very close to the captured one.•The use of the EKF as a controller is an efficient way to stabilize the model.

A new parameter identification method for a three-dimensional foot–ground contact model is presented. The model is used to reproduce the relationship between the contact forces and the relative foot–ground displacements and velocities. The parameters of the contact model are estimated using the optimization method known as covariance matrix adaptation evolution strategy. An extended Kalman filter is implemented as a controller to compute a forward dynamic analysis of the foot motion using body segment parameters and the ankle joint wrench as input data. The aim of this work is to adjust the position and size of the contact elements (spheres) and the model parameters in order to obtain both, a predicted motion provided by forward dynamics as faithful as possible to the captured motion and a resultant foot–ground wrench (obtained through the foot–ground contact model) as close as possible to the measured foot–ground reactions. The results show that the obtained motion is really similar to the captured one and, moreover, the vertical force and the moments in the horizontal plane are in agreement with the experimental mesurments. However, the bristle friction model used for tangential forces provides lower level of agreement with the experimental data.