Showing papers by "Yoshihito Amemiya published in 2001"

Quantum-Dot Logic Circuits Based on the Shared Binary-Decision Diagram

Abstract: We propose a method of constructing quantum-dot logic circuits that can be used to develop large digital systems with ultralow power consumption. These dot circuits consist of dot arrays fabricated using a wrap-gate structure, and they perform logic operations on the basis of the binary decision diagram. Sample dot circuits such as elementary logic gates and adder subsystems are designed. The operation of the designed circuits is confirmed by computer simulation.

Boltzmann machine neural network devices using single-electron tunnelling

Abstract: We proposed a method of implementing the Boltzmann machine neural network on electronic circuits by making use of the single-electron tunnelling phenomenon. The single-electron circuit shows stochastic behaviour in its operation because of the probabilistic nature of the electron tunnelling phenomenon. It can therefore be successfully used for implementing the stochastic neuron operation of the Boltzmann machine. The authors developed a single-electron neuron circuit that can produce the function required for the Boltzmann machine neuron. A method for constructing Boltzmann machine networks by combining the neuron circuits was also developed. The simulated-annealing operation can be performed easily by regulating an external control voltage for the network circuits. A sample network was designed that solves an instance of a combinatorial optimization problem. Computer simulation demonstrated that, through the simulated-annealing process, the sample network can converge to the global minimum energy state that represents the correct solution to the problem.

Analog computation using quantum-flux parametron devices

Abstract: Analog computation is a processing method that solves a given problem by utilizing an analogy of a physical system to the problem. An idea is presented here for relating the behavior of quantum-flux parametron circuits to analog computation. As an example, a method is proposed for solving a combinatorial optimization problem, the max-cut problem, by utilizing the properties of quantum-flux parametron circuits. In problem solving, a parametron circuit is constructed whose free energy is related to the objective function of a given problem and then is made to settle down to its minimum energy state. The solution to the problem can be obtained by checking the final state that the circuit reaches. The effectiveness of this method was confirmed by computer simulation for sample problem instances.

A νMOS Vision Chip Based on Cellular-Automaton Processing

Abstract: A neuron MOS (νMOS) vision chip was designed and fabricated for developing high-speed parallel image-processing systems based on cellular-automaton processing. The chip consists of cellular νMOS circuits that implement two fundamental functions in digital image processing: i) cleaning up noise in binary images and ii) detecting edges in the images, in addition to photosensing and image quantizing. Experimental results reveal that the fabricated chip successfully extracted edges from noisy inputs, which demonstrates the great potential of the νMOS vision chip in future image-processing applications.

A novel reaction-diffusion system based on minority-carrier transport in solid-state CMOS devices

Abstract: We propose a novel silicon device for imitating autocatalytic and dissipative phenomena of reaction-diffusion (RD) systems, using the minority carriers in semiconductors as diffusion substances. Thus, the diffusion of chemical substances in the RD system is imitated by that of the minority carriers. The chemical reaction, which results in the change of the concentration of the substances, is imitated by a reaction device. Numerical simulations show that the proposed RD device can successfully produce propagating waves in the same way as natural RD systems. Our results indicate that the proposed RD device will be a useful tool for developing novel hardware based on the RD mechanism.

Analog CMOS Implementation of Diffusive Lotka-Volterra Neural Networks

Abstract: Analog integrated circuits that implement oscillatory neural systems are very important tools for exploring and discovering novel forms of neural information processing. However, there are severe difficulties in constructing compact circuits with low-power dissipation, which prevents large-scale integration of oscillatory neural circuits. In this report, aiming at the development of oscillatory neural VLSIs, we present novel CMOS circuits that implement the Lotka-Volterra (LV) oscillatory systems.

Reaction-Diffusion Devices Using Minority-Carrier Transport in Semiconductors

Abstract: The purpose of this work is the development of novel silicon devices that imitate the lively, dynamic behavior of the reactiondiffusion (RD) system. The RD system is a complex system in which reaction and diffusion of chemical species coexist under a nonequilibrium condition[1]. It produces a variety of orders, rhythms and selforganizing phenomena observed in nature and in life. We are developing electronic analogues of the RD system with the aim of creating novel information-processing hardware. As the first step toward the goal, we propose in this report a RD device that utilizes autocatalytic multiplication and diffusion of minority carriers in a p-n-p-n diode array.

An Experimental Chip for Bio-Inspired Locomotion Controller based on the Wilson-Cowan Neural Oscillator

Abstract: Introduction We propose an analog CMOS circuit that implements a biologically-inspired locomotion controller for quadruped walking robots. The circuit is designed for emulating a central pattern generator (CPG), which is a biological neural network that can generate rhythmic movements for locomotion of animals [1]. We have fabricated a prototype chip using standard CMOS technology. From the experimental results on the prototype chip, we have confirmed desired performances of the proposed circuit. CPG Circuit The proposed circuit is based on a CPG model constructed from the Wilson-Cowan (WC) neural oscillator that can be easily implemented on a silicon chip [2]. Figure 1 shows the schematic of the WC neural oscillator circuit that consists of excitatory and inhibitory neuron circuits, both of which interact with each other via capacitive couplings. We constructed a network circuit (hereafter call the CPG circuit) from the WC neural oscillator circuits. Depending on coupling configurations (Fig. 2), a CPG circuit can produce various rhythmic patterns that correspond to the typical locomotion patterns of animals, such as walk, trot, and gallop. We have applied such rhythmic patterns to locomotion control in quadruped walking robots. Experimental results The CPG circuit was fabricated on a prototype chip using standard CMOS process (MOSIS AMIS 1.5-um). By changing circuit configurations, we measured voltages on the fabricated chip. Figures 3(a)-(b) show waveforms of the voltages that correspond to the locomotion patterns, such as trot and walk. The results show that the CPG circuit can produce stable rhythmic patterns under noisy environment (SNR:30dB). Conclusions An analog CMOS circuit was proposed for locomotion control in quadruped walking robots. By experiments on a fabricated chip, we have confirmed that the CPG circuit has capability of producing stable rhythmic patterns under an actual (noisy) environment. Such characteristics of the CPG circuit are suitable for robot locomotion control in real situations. In future work, we are going to incorporate our chip into a quadruped walking robot.