An indirect stiffness estimation of flexible joint based on system function approximation

We propose an indirect stiffness estimation method of flexible joint based on system function approximation in this paper. At first, an alternative system function is constructed. Then a system function estimation method based on saturation function is introduced to estimate its output. Taking the estimated system function as prior model, the least square method is finally introduced to identify the joint stiffness. Meanwhile, another system estimation method based on RBF neural network is also introduced to verify the efficiency of the former one. There are two advantages of the new proposed stiffness estimation method. Firstly, it is more robust to environmental noise, since no basic version or filter version of angle acceleration is maintained in the estimation process. Secondly, it's easy to apply because only measurements of joint angle on motor side, joint velocity on motor side and joint angle on link side is needed.