چكيده به لاتين
Designing a stable walking gait for biped robots with point-feet is stated as a constrained nonlinear optimization problem which is normally solved by an offline numerical optimization method. On the result of an unknown modeling error or environment change, the designed gait may be ineffective and an online gait improvement is impossible after the optimization. In this thesis, we apply Generalized Path Integral Stochastic Optimal Control to closed-loop model of planar biped robots with point-feet which leads to an online Reinforcement Learning algorithm to design the walking gait. The results show that the proposed algorithm is very successful to adapt the controller of Rabbit, which is a planar biped robot with point-feet, for stable walking with desired features. We have continued with designing a robust stable walking gait for biped robots against a known range of disturbances, which is very important in real applications. Since, the gait designed by the proposed algorithm might not be robust enough against disturbances, we extend a robust version of the proposed algorithm to design an exponentially stable walking gait which is robust against modeling errors/disturbances. It is done by minimizing the costs of worst rollouts which are generated in presence of different modeling errors/disturbances. Time-invariant controllers generally guaranty the stability of a biped robot with point-feet which is a very interesting feature. However, complex hybrid dynamics of quadruped robots made designing the time-invariant controller very difficult. Therefore instead of designing a unique time-invariant controller for a quadruped robot, we decompose the robot into two biped robots which are controlled by two time-invariant controllers simultaneously. Then we introduce how to extend the proposed algorithm to adjust the parameters of the two controllers. The results show that using the extended algorithm, an stable walking including the desired features is attained for a new situation and the modeling error is quickly compensated.
