Haichao Liang

Gabor features for real-time road environment classification

In this paper we present a methodology to use Gabor response features for real-time visual road environment classification. Processing Gabor filters using hardware solely dedicated to this task enables improved real-time texture classification. Using such hardware enables us to successfully extract Gabor feature information for a four-class road environment classification problem. We used summary histogram as an intermediate level of texture representation prior to final classification. Overall we obtain a maximally correct classification circa 98%, outperforming prior work in the field.

An FPGA-based CollisionWarning System Using Hybrid Approach

In this paper, we propose an FPGA-based collision warning system for advanced automobile driver assistance systems or autonomous moving robots. The system consists of three function blocks: coarse edge detection using a resistive-fuse network, moving-object detection inspired by neuronal propagation in the hippocampus, and danger evaluation and collision warning using fuzzy inference. The first two functions are implemented in FPGAs. The system can detect moving objects with a speed range of 3-192 km/h with a sampling period of 30 ms for an input image of 320 x 256 pixels, and can output a warning against dangerous regions in the input image.

Silicon nanodisk array with a fin field-effect transistor for time-domain weighted sum calculation toward massively parallel spiking neural networks

A nanodisk array connected with a fin field-effect transistor is fabricated and analyzed for spiking neural network applications. This nanodevice performs weighted sums in the time domain using rising slopes of responses triggered by input spike pulses. The nanodisk arrays, which act as a resistance of several giga-ohms, are fabricated using a self-assembly bio-nano-template technique. Weighted sums are achieved with an energy dissipation on the order of 1 fJ, where the number of inputs can be more than one hundred. This amount of energy is several orders of magnitude lower than that of conventional digital processors.

A 2-dimensional Si nanodisk array structure for spiking neuron models

Spiking neuron models, which simplify the biological neuron function, have attracted much attention recently in the fields of computational neuroscience and artificial neural networks. In these models, generation of post-synaptic potentials (PSPs) is an essential function. In this paper, we propose a new nanodevice structure using a nanodisk array connected to a MOSFET for spiking neuron models. The structure generates PSPs by taking advantage of the delay in electron hopping movement among nanodisks. The results of single-electron circuit simulation demonstrate the controllability of PSP shapes by a control gate placed over the nanodisk array.

Digital VLSI Implementation of a Moving Object Detection Algorithm Based on Neuronal Propagation in the Hippocampus

In this paper, we propose a moving object detection algorithm for the advanced driver assistance system (ADAS) and its digital VLSI implementation. The algorithm is based on parallel processing inspired by neuronal propagation underlying sequence coding in a model of the hippocampus. We have implemented this algorithm on an FPGA, and have verified the desired operation: detection of the velocity and direction of moving objects at real time

Spike-based time-domain weighted-sum calculation using nanodevices for low power operation

This paper introduces a time-domain weighted-sum calculation operation based on a spiking neuron model, and discusses a resistance-capacitance circuit that performs a calculation operation assumed to be realized in CMOS VLSI technology. A nanodevice that executes this calculation is also presented. The calculation circuit is useful for extremely low power operation. This operation uses the rising slopes of post-synaptic potentials triggered by input spike pulses. In the time-domain calculation circuit, the energy dissipation is independent of the resistance, and only depends on the capacitance and voltages. However, the time constant, which is the product of the resistance and capacitance, should be relatively large to guarantee the calculation resolution, and therefore the resistance should be at the giga-ohms levels. The nanodevice consists of a nanodisk array connected with a fin field-effect transistor. Nanodisk arrays can be fabricated using a self-assembly bio-nano-template technique, and they act as resistors with resistance levels of several giga-ohms. A weighted sum can be achieved with an energy dissipation on the order of 1 fJ, with a number of inputs that can be more than 100. This amount of energy is several orders of magnitude lower than that of conventional digital processors.

A silicon nanodisk array structure realizing synaptic response of spiking neuron models with noise

In the implementation of spiking neuron models, which can achieve realistic neuron operation, generation of post-synaptic potentials (PSPs) is an essential function. We have already proposed a new nanodisk array structure for generating PSPs using delay in electron hopping among nanodisks. Generated PSPs have fluctuation caused by stochastic electron movement. Noise or fluctuation is effectively used in neural processing. In this paper, we review our proposed structure and show fluctuation controllability based on single-electron circuit simulation.

A motion detection model inspired by hippocampal function and its applications to obstacle detection

We have proposed a motion detection model, CA3-GU-CA1 (CGC) model, inspired by hippocampal function. The CGC model treats edges extracted from monocular image sequences, and detects motion of the edges on segmented 2D maps without image matching. In this paper, we propose an FPGA implementation of the CGC model, in order to achieve low power processing toward practical use. Then, we propose an obstacle detection algorithm using time-to-collision (TTC) based edge grouping. We have evaluated the performance of motion and obstacle detection by using artificial and real image sequences. The results show that the CGC model can achieve high detection rate in complicated situations, and can achieve accurate detection when using a high frame-rate. The proposed obstacle-detection algorithm can detect dangerous objects moving across based on a novel TTC estimation algorithm. Both motion detection and obstacle detection parts can operate at more than 1000fps. The CGC model can also operate with a power dissipation of about 1.4W based on the FPGA implementation.

Parametric Control in a Region-Based Coupled MRF Model with Phase Dynamics for Coarse Image Region Segmentation

Towards hardware implementation of real-time visual image processing, we propose a piecewise linear (PWL) approximation of a coupled region-based Markov Random Field (MRF) model with hidden phase variables for coarse image region segmentation. We introduce PWL functions into update equations of the region-based model, in order to make it easy to control parameters that determine the balance between image segmentation and smoothing, as well as to make it efficient for hardware implementation. We can tune filtering properties of our model by controlling the parameters of the PWL functions in applications to a task of coarse image region segmentation. Finally, we demonstrate that closed regions in input images can be represented by phase variables in multi-scale.

An FPGA-Based Collision Warning System Using Moving-Object Detection Inspired by Neuronal Propagation in the Hippocampus

In this paper, we propose an FPGA-based collision warning system for advanced automobile driver assistance systems or autonomous moving robots. The system consists of three function blocks: edge detection, moving-object detection and danger evaluation and collision warning. In the moving-object detection, the system uses a moving-object detection algorithm inspired by neuronal propagation in the hippocampus, which can run in high speed and low calculation cost. We have applied the system in a robot. It can detect moving objects with a speed range of 3-47cm/s with a sampling period of 33ms for an input image of 320×240 pixels, and can output a warning against dangerous regions in the input image.