Pneumatic anti-vibration apparatus (AVA) has been used for suppressing the vibration from the floor. Compressed air produced by the air compressor is supplied to air spring, which is used as an actuator in AVA. This paper considers the suppression of flow disturbance, which is caused by pressure variation of compressed air. Based on the internal model principle, the suppression of vibration of AVA due to flow disturbance has been confirmed by using Central Pattern Generator (CPG) in parallel with a displacement controller. In biological study, CPG in bodies playing a significant role in the walking have been learned. Moreover, mathematical models of CPG have been reported, and some researchers have been studying about walking of biped-robots by using CPG to generate desired value. On the other hand, this paper proposes asymmetric CPG as new model for further vibration suppression, compared with the symmetric CPG which is the conventional model. This new model introduces asymmetry by changing its parameters into different values in order to improve reproducibility of waveform which shows the variation of supplied air pressure. Validity of the proposed model is presented by experimental result of AVA. The effectiveness is shown by analyzing and comparing the results of experiment and simulation of symmetric and asymmetric CPGs.

/pdf/control-for-flow-disturbance-suppression-of-pneumatic-anti-9gyi1i0q3y.pdf

Control for flow disturbance suppression of pneumatic anti-vibration apparatus using asymmetric central pattern generator

Pneumatic anti-vibration apparatus (AVA) has been used for suppression of vibration from floor in the field like semiconductor manufacturing. The compressed air supplied to the pneumatic AVA is generated by the air compressor. During the compression process of the air, the pressure of the compressed air varies. Consequently, the pressure variation changes the flow rate of the air supplied to the pneumatic AVA, which is called flow disturbance. Based on the internal model principle, the suppression of vibration of AVA due to flow disturbance has been confirmed by using Central Pattern Generator (CPG). Previous studies in the field of biology have confirmed the existence of a neuronal network that generates a walking rhythm in the spinal cord of an organism, and this is called CPG. Some researchers have applied CPG to such as walking of biped-robots. However, there are no applications using information inside CPG. This letter proposes using internal information in CPG as new method for further vibration suppression, compared with the conventional method. Specifically, this method makes use of outputs corresponding to differential and integral signals inside the CPG. It is shown the effectiveness of the proposed method by experimental results.

A trial of flow disturbance suppression for pneumatic anti-vibration apparatus using internal information in central pattern generator

A neuro-oscillator system is modeled as a hybrid system containing binary switches. In case of excitatory connected system, oscillations of neuron models synchronize mutually. We give a theoretical interpretation to these synchronizations by using a transformation into a decoupled system. The amplitude and frequency of the limit cycle oscillation are obtained as approximated functions and it is shown how parameters of system affect the oscillations. In numerical simulations, the number of mutually exciting neurons modulates the strength of attractions of the limit cycle.

http://ieeexplore.ieee.org/iel5/9240/29285/01323357.pdf

Synchronized oscillations of neuro-oscillator systems with mutually exciting connections

空圧式除振装置に対するcpg適用の留意点（cpgのパラメータ設計，スイッチ入力の変更，そして反転・非反転cpgの比較）

As a central pattern generator, a neuro-oscillator with four neurons was used for the motion generation of a biped locomotive robot. There is a problem to determine unknown parameters in nonlinear differential equations of the neuro-oscillator so that it generates joint trajectories which yields reasonable biped locomotion. For this problem, we apply the describing function method for the neuro-oscillators, then we obtain equations which should be satisfied for the existence of a periodic solution. Adding a walking pattern which is represented by a trajectories of swinging leg, we can determine the unknown parameters from the equations obtained above. The effectiveness of the proposed method as a design procedure of the biped locomotion was assured by simulation study and experiments with a small biped robot.

https://www.jstage.jst.go.jp/article/sicetr1965/34/7/34_7_830/_pdf

Osamu Katayama

Papers

Application of the Describing Function Method to Motion Generation Using a Neuro Oscillator

Neural Networks and Learning of Robots