Masahiro Ohtani

Analog MOS Circuits for Motion Detection Based on Correlation Neural Networks

We propose simple analog MOS circuits producing one-dimensional optical flows aiming at the realization of compact motion-sensing circuits. In the proposed circuit, the optical flow is computed by a number of local motion sensors which are based on biological motion detectors. Mimicking the structure of biological motion detectors made the circuit structure quite simple, compared with conventional velocity sensing circuits. Extensive simulation results by a simulation program of integrated circuit emphasis (SPICE) indicated that the proposed circuits could compute local velocities of a moving light spot and showed direction selectivity for the moving spot, which implies that a high-resolution motion-sensing chip can be realized using standard analog very large-scale integration (VLSI) technology.

Analog Metal-Oxide-Silicon IC Implementation of Motion-Detection Network Based on a Biological Correlation Model

In this paper, fabrication of a simple analog network for one-dimensional motion detection using the complementary metal-oxide-silicon (CMOS) process is described. The network is constructed with a number of local motion sensors based on biological elementary motion detectors. Experimental and simulation results indicate that the fabricated network computes the local velocity and direction of the moving object in the velocity range of 10-4 m/s to 3.2 ×102 m/s. The results also show that the velocity sensitivity of the network can be changed flexibly. The circuit structure of the network is compact and suitable for highly dense motion sensor arrays.

Analog velocity sensing circuits based on bio-inspired correlation neural networks

We propose simple analog MOS circuits producing one-dimensional optical flows aiming at the realization of compact motion-sensing circuits. In the proposed circuit, the optical flow is computed by a number of local motion sensors, as in biological motion-sensing systems. Mimicking the structure of biological motion detectors made the circuit structure quite simple, compared with conventional velocity sensing circuits. Extensive simulation results using SPICE and experimental results indicated that the proposed circuits could compute local velocities of a moving light spot and showed direction selectivity for the moving spot, which implies that a high-resolution motion-sensing chip can be realized using standard analog very large-scale integration technology.

Analog LSI Implementation of Biological Direction-Selective Neurons

A direction-selective motion-detection network is implemented into analog metal-oxide-semiconductor (MOS) circuits. The network, based on biological direction-selective (DS) neurons, was constructed with three functional components that act as an elementary motion detector, a dendritic tree and a cell body in a DS neuron. The elementary motion detector and dendritic tree were realized using a novel current-mode delay line circuit. Since the delay line circuit utilized a Miller capacitance caused by the Miller effect and a parasitic capacitance, the occupied area of the network on a LSI chip was reduced effectively. Therefore, it is possible to reduce an occupied area of the network effectively in a LSI chip. Simulation results using the simulation program with integrated circuit emphasis (SPICE) showed that the network operated with the directional selectivity of an object motion. A two-dimensional direction selective network was designed for a LSI, which detects the two-dimensional motion of an object. This simple network has the potential to enable forming motion-detection systems without arithmetic calculation.

Analog MOS circuit systems performing the visual tracking with bio-inspired simple networks

In this paper we propose simple neural networks for the visual-tracking task which are inspired from biological eye-movement systems. Although the proposed system is very simple, the system can pursue the visual objects in the presence of background images. The proposed system consists of velocity sensing circuits, competitive neural circuits and motor control circuits which are contrived to be suitable for a large-scale analog circuit implementation. Results of extensive numerical and SPICE simulations indicate that the proposed circuit can produce the appropriate motor signal for moving visual objects and pursue the object in the presence of background images.

An analog integrated circuit for motion detection

It was clarified in motion detection that the analog network based on a biological system works in a similar manner to a simplified algorithm of the optical flow. The proposed network is constructed by combining simple fundamental circuits called VSC. The direction and speed of motion can be detected with the network. The network is expandable for large scale integration. A motion detection system for the two dimensional space was proposed based on biological systems.

Elemental devices for monolithic optoelectronic integrated circuits on lattice-matched Si/III-V-N/Si structure

Optoelectronic integrated circuits (OEICs) have been expected for realization of novel devices and circuits. It is necessary for realization of OEICs to combine Ill-V compounds and Si with high quality. However, many dislocations were induced by the large difference in valence electron and lattice-constant between Ill-V compounds, such as GaAs and InP, and Si. We have already realized a dislocation-free and lattice-matched Si/GaPN/Si using molecular beam epitaxy (MBE)[1]. In addition, a dislocation-free InGaPN/GaPN double hetero (DH) light emitting diode (LED) has been realized on a Si substrate[2]. If these technologies are merged systematically, the top Si layer and III-V-N alloys will be used for integrated circuits and optical devices, respectively, as shown in Fig. 1 [3]. In this study, we have investigated a fabrication process for monolithic OEICs based on Si/III-V-N/Si structure and realized MOSFETs and LEDs in the same Si/III-V-N/Si wafer. A Si/III-V-N/Si wafer was grown by MBE, as shown in Fig. 2. A p-Si(100) substrate misoriented 40 toward [011] direction with carrier concentration of 5x1 OI8cm3 was used as a substrate. A GaP layer grown by migration enhanced epitaxy (MEE) is necessary to solve the difference in valence electron between Si and Ill-V compound semiconductors. An InyGa1-yPO.96No.04/GaPo.98No.o2 DH structure was grown at 500°C in the III-V-growth chamber. The GaP098N002 layers are lattice matched to Si. InyGa1 yP0.96No.04 layer is an active layer for DH LEDs. After the growth of DH structure, a thermal treatment was performed at 450°C and the sample was transferred to the Si-growth chamber through a vacuum chamber. A 1 ptm-thick Si layer was grown by electron-beam evaporator on the n-GaPN layer. A carrier concentration in the Si epilayer of 4x1017cm-3 was measured by four-point probe method. A fabrication sequence is proposed in Fig. 3. A 1ptm-thick field-oxide layer was deposited by chemical vapor phase deposition after separating LED regions by reactive ion etching. For simplicity of the process, a thermal annealing to increase the luminescence efficiency and a post-annealing of ion-implantation were combined with a gate-oxidation process. The luminescence of GaPN layers has been increased by the annealing at 900°C for 1-10min[4]. Thus, gate-oxidation was performed at 900°C for 10min in a wet 02 ambient, since a growth rate is higher than that in dry oxidation. Figure 4 shows a micrograph of a chip fabricated through the OEIC process. pMOSFETs and LEDs are shown in the same chip. Capacitance-voltage characteristics of MOS diodes showed surface inversion. The thickness of gate oxide was estimated at 16nm, which was close to a thickness of oxide grown on a bare n-Si wafer as a reference. The carrier concentration was estimated at 6.6x1 017cm-3. The wet oxidation was found to be utilized for the OEIC process. Figure 5(a) shows typical characteristics of drain currents(IdS) versus drain voltages(Vd,). Although a threshold voltage of -3.1V was relatively high, the pMOSFET characteristics have been obtained on the OEIC wafer for the first time. It is necessary to improve the growth process for reducing the threshold voltage and carrier concentration in Si epilayer, which could be caused by incorporation of P atoms during the growth. The InGaPN/GaPN DH LEDs in Fig. 4 showed electroluminescence at room temperature, as shown in Fig.5(b). Red luminescence of 2Ox45 im2 LED was also found by visual observation even at a current of 800tA. The properties ofLEDs will be investigated in detail. In conclusion, we have realized the elemental devices for monolithic OEICs on the lattice-matched Si/III-V-N/Si structures by the proposed OEIC process. This work was partially supported by the Ministry of Education, Science, Sports and Culture (Specially Promoted Research and the 21st Century COE Program).

Analog network for detection of approaching object with shape recognition based on visual systems of lower animals

We propose an analog two-dimensional network for detection of an approaching object with simple shape recognition based on the visual systems of lower animals such as locusts and frogs. It was shown with the simulation program with integrated circuit emphasis (SPICE), that a proposed network constructed with simple analog circuits can detect the two-dimensional direction of an approaching object and simultaneously recognize the simple shape such as a circle, square, triangle and rectangle. The application of the network is expected to realize collision avoidance.

An Analog Integrated Circuit for Motion Detection Based on Biological Correlation Model

The ability to perceive moving object is an important visual modality. An optical flow, which represents a local velocity field caused by the moving objects, has been studied in the research field of computer vision[1]. The optical flow chip has been tried to be realized by J. Kramer et al[2]. We have tried to simplify the fundamental analog integrated circuit for motion detection combining biological systems with a simplified algorithm of the optical flow. In this paper, we show experimental results of a fabricated chip which contains unit circuits for the motion detection. Results of SPICE simulations, based on those results and the simplified algorithm, showed that the proposed circuit can detect the direction and speed of motion.