The specific scheme of artificial sensory nervous system based on memristor proposed by microelectronics

At present, human society is evolving from informatization to intelligence. An intelligent society requires an efficient and intelligent information perception system to effectively identify, process and make decisions on the huge amount of information perceived, and to effectively filter repetitive and meaningless information. Therefore, it will become an important development trend to construct an efficient intelligent information perception system with biological reality based on the functional characteristics of the biological sensory nervous system.

Recently, the team of Academician Liu Ming from the Key Laboratory of Microelectronics of the Institute of Microelectronics, Chinese Academy of Sciences has proposed an implementation scheme of an artificial sensory nervous system based on the habituation characteristics of memristors. Habitual spiking neural network for autonomous cruise obstacle avoidance.

The research team constructed a sensory neuron based on Mott memristors and sensors. The neuron can sense external analog signals and convert them into real-time dynamic pulse signals, realizing the basic function of sensing and transmitting external signals. The sensory neurons are further connected with the relay neurons through synaptic devices to construct the habituation perception system.

The synaptic device has the habituation evolution trend of the weight under continuous stimulation, which in turn affects the transmission efficiency of the sensory neuron signal to the relay neuron, so that the output of the relay neuron exhibits a frequency drop characteristic (that is, the habituation characteristic, as shown in Fig. a). Based on this habituation feature, the team further constructed a habituation spiking neural network to realize the robot obstacle avoidance function.

The test results show that the memristor-based artificial sensory neural system constructed based on habitual learning rules can effectively improve the robot’s obstacle avoidance efficiency.

The habituation sensory nervous system can also be applied to different perceptual systems, such as smell, taste, vision, hearing, etc., through different sensors. By realizing the perception system of biological reality, it is expected to realize a more biologically intelligent terminal system.

a, Memristor-based habituation sensory nervous system schematic diagram and system response characteristics. b. Verification of Memristor-based Habitual Pulse Neural Network in Improving the Efficiency of Robotic Obstacle Avoidance

The Links:   SKKH330/16E FP25R12KT3 PM100RSE120

Related Posts