Bio-heuristic curiosity cognitive development system and operating method thereof

Abstract

The invention relates to a bio-heuristic curiosity cognitive development system and an operating method thereof, belonging to the technical field of intelligent robots. The bio-heuristic curiosity cognitive development system is constructed by simulating the coordination mechanism of basal ganglia, cingulate cortex and the regulation mechanism of persistent dopamine, including a perceptual part, an evaluative part, a behavioral part, a value decreasing part, a curiosity part and a persistent dopamine regulation part. The invention provides an object-heuristic curiosity cognitive development system, and the problem of poor autonomous exploratory ability of robot in unknown environment is solved. Secondly, the mechanism of value attenuation is introduced, which can make the learned knowledgecontinuously affect the subsequent learning process of robot, make the robot reach the goal quickly, and solve the problem of low learning efficiency of traditional reinforcement learning.

Description

Technical field [0001] The invention relates to a biologically inspired curiosity cognitive development system and an operating method thereof, and belongs to the technical field of intelligent robots. Background technique [0002] Cognitive development is that humans acquire knowledge and skills autonomously through interaction with the environment. The human brain system plays a very important role in the process of human cognitive development. It simulates the structure and coordination mechanism of the human brain and gives it to robots. It is artificial An important research topic of intelligence and control science. [0003] In 1996, J. Weng first proposed the idea of ​​autonomous mental development of robots. He believed that agents should be based on simulating the human brain and under the control of internal development programs to develop mental abilities through the interaction of sensors and effectors with the unknown environment. The first stage of the cognitive deve...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to solve the problems of traditional robot continuous behavior learning self-adaptation and poor self-learning ability. Based on the neurophysiology and neuropsychology research related to the human brain, based on the Actor-Critic model, the curiosity exploration mechanism and value are introduced. The attenuation mechanism uses a recurrent neural network to simulate the information processing mechanism in the human brain, so that the robot can gradually learn to balance control skills by autonomously exploring the behavior space in an unknown environment, thus providing a biologically inspired curiosity and cognitive development system and method of operation

Images