predictive model

In this paper, the Nonlinear Model Predictive Control (NMPC) approach is employed to control the five-DOF biped robot in moving up and down the stairs. In the proposed method, the path planning is performed in real time. That is, there are no predefined trajectories to track by the robot and the motion of the robot is expressed in terms of some constraints. The main feature of the proposed method is the online path planning, which leads to more robustness of the controller against the uncertainties in the environment. Moreover, the NMPC is designed in such a way the robot is able to climb the stairs with different heights and depths; hence, the gaits are defined adaptively. One major issue in using the MPC is its dependency to an accurate model. In this paper, the robot is modeled online using the Radial Basis Function (RBF) neural networks. In this way, the NMPC is turned into an adaptive controller that can adapt itself to changes in the system parameters. Simulation results show good performance of the proposed method as compared to the recently proposed methods in literature.