论文信息 - Vibration suppression control using a pattern generator for a robot driven by air actuators

Vibration suppression control using a pattern generator for a robot driven by air actuators

For a communication robot like an android, air actuators are preferable because their elastic characteristics would reduce damages to the human or furnitures. However, due to the low controllability of the actuators and the complex structure, the robot makes undesirable motions such as vibrations. In order to overcome this problem, we propose a control method using a pattern generator and a reinforcement learning algorithm. We applied our method to the control of a 3-DOF robot, and the experimental result in simulation showed that undesirable vibrations can be suppressed by our method.

[1] Kenji Doya,et al. Adaptive Synchronization of Neural and Physical Oscillators , 1991, NIPS.

[2] Andrew G. Barto,et al. Reinforcement learning , 1998 .

[3] Shigenobu Kobayashi,et al. An Analysis of Actor/Critic Algorithms Using Eligibility Traces: Reinforcement Learning with Imperfect Value Function , 1998, ICML.

[4] Hiroshi Shimizu,et al. Self-organized control of bipedal locomotion by neural oscillators in unpredictable environment , 1991, Biological Cybernetics.

[5] Shin Ishii,et al. Reinforcement learning for a biped robot based on a CPG-actor-critic method , 2007, Neural Networks.

[6] Minoru Asada,et al. CB2: A child robot with biomimetic body for cognitive developmental robotics , 2007, 2007 7th IEEE-RAS International Conference on Humanoid Robots.