Abstract:

Virtual impedance control can be used to avoid obstacles, joint limits, and self-collisions. In this thesis, we propose a stabilizer that uses virtual impedance control and maximal output admissible (MOA) sets to guarantee balance stability and execute other functions simultaneously while walking. This stabilizer ensures that the zero moment point (ZMP) does not exceed the constraints by changing the gains of virtual impedance control after computing the MOA sets so that the humanoid robot will not fall.

Furthermore, we use the gradient projection method to control the motion of the humanoid robot in the null space of virtual impedance control. This enables us to execute subtasks while the main tasks are completed through different cost functions.