基于忆阻的嗅觉感知及跨模态联想电路设计与实现

Memristive circuit design and implementation of olfactory perception and cross-modal association

嗅觉研究在环境监测、灾害预警等领域的重要性凸显,联想记忆是生物大脑处理信息的机制,探索嗅觉感知及其跨模态联想对于扩展类脑感知在机器人技术上的应用和发展十分必要.本文基于忆阻设计了嗅觉感知与跨模态联想电路,模拟了嗅觉的基本功能及其情感、记忆效应,主要包括3个模块:气味识别模块以忆阻嗅觉神经元为基础实现对两种气味的识别;高级感知模块搭建情感产生电路生成嗅觉情感,并利用忆阻阻值变化实现对其他模态记忆巩固的影响;延时跨模态联想模块基于忆阻变化规律设计延时电路和联想记忆电路,以情感为强化物实现了嗅觉与其他模态信息的延时联想.最后设计4种情景进行实验,仿真结果表明该电路可以实现嗅觉的感知功能和与其他模态信息的跨模态联想,为机器人嗅觉感知提升提供了可能.

The importance of olfactory research in environmental monitoring,disaster warning,and other fields is highlighted.Associative memory is the mechanism by which biological brains process information.Exploring olfactory perception and its cross-modal association is necessary to expand the application and development of brain-like perception in robotics technology.This paper designs olfactory perception and cross-modal association circuits based on memristor,simulating the basic function of olfaction and its emotion and memory effects,which mainly includes three modules:an odor recognition module realizes recognizing two types of odors based on memristor olfactory neurons;an advanced perception module builds an emotion generation circuit to generate olfactory emotions,and uses changes in memristance to impact the consolidation progress of other modal memories;and the delayed cross-modal association module designs a delay circuit and an associative memory circuit based on the variation law of memristance and implements delayed association among olfactory information and other modal information using emotions as reinforcement.Finally,four cases are designed for experiments,and the simulation results show that the circuit can achieve olfactory perception function and crossmodal association with other modal information,allowing for improving a robot’s olfactory perception.