Artificial Visual Neural System Based On Optoelectronic Memristor

More than 80%of human information is obtained through visual information,and the development of artificial vision systems capable of sensing optical information plays a crucial role in liberating production and increasing machine efficiency.For the current artificial vision system,it is usually composed of photoreceptors,information memory units and processing units that perform complex image processing tasks based on complementary metal oxide semiconductor（CMOS）units.Although the system can continuously detect images in real time,the lack of biological correlation in this system leads to the inability to preprocess data and generates a large amount of data redundancy,it consumes a lot of storage space and leads to high power consumption.In contrast,the sensory neurons in the retina of the human visual system can not only detect light stimulation,but also perform the first stage of image processing before the more complex visual signal processing in the human visual cortex.Therefore,using a new device with biological similarity,memristors,to construct a visual system with biological characteristics is very important in improving computational efficiency and reducing system power consumption.According to the biological vision system,two key elements must be developed to construct the artificial vision system:artificial optical synapses and neurons.The biological synapse is the connection site between two neurons.It releases Ca²⁺and Na⁺neurotransmitters into the synaptic gap through the presynaptic membrane,thus changing the weight of the synapse,making the connection strength between the two neurons change,thus realizing the storage and calculation of information.Optical synapse not only has the function of biological synapse,its weight is also regulated by external optical signals,and it can process data while sensing external information.Neurons can accumulate information in synapses and convert it into corresponding chemical signals or electrical signals.When the information reaches a certain threshold,it will be transmitted according to certain coding rules.However,there is little research on artificial visual neural systems based on memristors.Most work units have studied the characteristics of basic devices,and there are problems such as low light response current,poor device stability,lack of biological wearability,and the need for further exploration of neural circuit models.In order to explore the artificial visual nervous system based on memristors,this paper respectively explores the optical synaptic devices,memristor neuron circuits,and builts the artificial visual nervous system based on full memristors to realize the function and behavior of bio-like neurons.The main research contents are as follows:1.Optical synaptic memristor:In this study,through efficient packing of individual CDs in two-dimensions,the synthesis of flexible CDs ribbons is demonstrated for the first time.Due to their flexibility and regular2D morphology,the CDs ribbons offer outstanding performance as the active layer material in transparent flexible memristors,with the developed devices providing excellent data storage,retention capabilities,and fast optoelectronic responses.A memristor device with a thickness of 8μm shows good data retention capability even after 10⁴ cycles of bending.Furthermore,the device functions effectively as a neuromorphic computing system with integrated storage and computation capabilities,with the response speed of the device being less than 5.5 ns.These properties create an optoelectronic memristor with rapid Chinese character learning capability.This work lays the foundation for wearable artificial intelligence.2.The neuron circuit based memristor:We propose an efficient and compact neuron of negative differential resistance（NDR）memristor based on Al As/In_0.8Ga_0.2As/Al As quantum well（QW）structure,which implements the integration feature of the neuronal membrane and avoids using external capacitors and successfully applies it to the self-designed super reduced neuron circuit.In addition,the NDR neuron memristor exhibits other comprehensive merits,such as low variation（0.264%）,high temperature resistance（400℃）,high endurance(more than 10¹¹)and neuromorphic computing.Moreover,the negative differential effect in NDR memristor emulates the action potential behavior of neurons,saving the hardware cost of neuron circuits,which only needs two NDR memristor and one inductor to realize rich neuron dynamics and Fitz-Hugh Nagumo（FN）biological neuron dynamics.This work opens the way for simulating FN neurons with QW based NDR memristors,and provides a more competitive method for building a highly integrated neural morphology hardware system.3.Artificial visual perception nervous system:a)We propose a fully memristor-based artificial visual perception nervous system（AVPNS）which consists of a quantum-dot-based optoelectronic memristor and a nanosheet-based threshold-switching（TS）memristor.We use an optoelectronic and a TS memristor to implement the synapse and leaky integrate-and-fire（LIF）neuron functions,respectively.With the proposed AVPNS we successfully demonstrate the biological image perception,integration and fire,as well as the bio-sensitization process.Furthermore,the self-regulation process of a speed meeting control system in driverless automobiles can be accurately and conceptually emulated by this system.Our work shows that the functions of the biological visual nervous system may be systematically emulated by memristor-based hardware system,thus expanding the spectrum of memristor applications in artificial intelligence.b)We developed a high-speed multifunctional artificial vision system capable of recognizing,memorizing,and actuating self-protection by combining a Sb₂Se₃/Cd S-core/shell（SC）nanorod array optoelectronic memristor,a threshold-switching memristor,and a mechanical eye.Light absorption and charge carrier extraction are advantages of optoelectronic memristors with high-quality SC nanorod arrays.The device achieves a fast response speed and a large response current of up to 40μs and 0.8μA.When the received light intensity of the optoelectronic memristor exceeds a preset range,it can make the mechanical eye move,simulate eye muscle contraction,and reproduce the self-protective reaction of closing the eyes when the human eye is injured by strong light.Artificial vision systems offer a potential technique for bio-nanotechnology,particularly in the domain of artificial intelligence simulation of biosensor systems.