Untanpreeda presented a training algorithm based on BP [1] , which guarantees the closed loop stability for a class of widely used Neural network control systems. However, it has some shortcomings, such as insuf...Untanpreeda presented a training algorithm based on BP [1] , which guarantees the closed loop stability for a class of widely used Neural network control systems. However, it has some shortcomings, such as insufficient stable condition, low efficiency and frequent convergence of parameters to a local minimum. A new training algorithm based on Alopex is proposed to ensure sufficient stability, and overcome some of the shortcomings.展开更多
Natural dynamics can be exploited in the control of biped walking robots: the swing leg can swing passively, the compliant ankle can naturally transfer the center of pressure along the foot and help in toe off. These ...Natural dynamics can be exploited in the control of biped walking robots: the swing leg can swing passively, the compliant ankle can naturally transfer the center of pressure along the foot and help in toe off. These mechanisms simplify control and result in motion that is smooth and natural looking. Imitated human being’s behavior, we presented a control strategy for HIT III biped robot, in which active and passive motions were combined. The experiment result shows that robot walked at a faster speed (approximately 0.25m/s) than previous and had a natural and smooth looking gait.展开更多
文摘Untanpreeda presented a training algorithm based on BP [1] , which guarantees the closed loop stability for a class of widely used Neural network control systems. However, it has some shortcomings, such as insufficient stable condition, low efficiency and frequent convergence of parameters to a local minimum. A new training algorithm based on Alopex is proposed to ensure sufficient stability, and overcome some of the shortcomings.
文摘Natural dynamics can be exploited in the control of biped walking robots: the swing leg can swing passively, the compliant ankle can naturally transfer the center of pressure along the foot and help in toe off. These mechanisms simplify control and result in motion that is smooth and natural looking. Imitated human being’s behavior, we presented a control strategy for HIT III biped robot, in which active and passive motions were combined. The experiment result shows that robot walked at a faster speed (approximately 0.25m/s) than previous and had a natural and smooth looking gait.
