Multi-brain area cooperative autonomous decision making method based on multi-modal fusion

Abstract

The invention belongs to the cognitive nerve technology field and especially relates to a multi-brain area cooperative autonomous decision making method based on multi-modal fusion. Problems that thecost of an existing unmanned aerial vehicle obstacle avoidance technology is high; the technology is not flexible; and an existing reinforcement learning method requires a control object to have a high fault tolerance capability are solved. The multi-brain area cooperative autonomous decision making method based on multi-modal fusion comprises the following steps of acquiring the space position information of an obstacle and inputting into a multi-brain area cooperative reinforcement learning model which is constructed in advance; and according to the reward information fed back by an environment, through dopamine regulation and control and a synaptic plasticity mechanism, updating the multi-brain area cooperative reinforcement learning model, and realizing unmanned aerial vehicle autonomous obstacle avoidance. In the invention, the dangerous degree of the obstacle in a scene can be accurately assessed, a brain autonomous learning process is simulated, the unmanned aerial vehicle can rapidly and accurately learn an obstacle avoidance strategy, autonomous obstacle avoidance is realized and a task is completed.

Description

technical field [0001] The invention belongs to the field of cognitive neurotechnology, and in particular relates to a multi-brain-region cooperative autonomous decision-making method based on multi-modal fusion. Background technique [0002] With the development of UAV technology, UAVs are widely used in many technical fields due to their practicability, and the active safety of UAVs is the basis for their safe application in real scenarios. The active safety of UAV refers to its ability to accurately perceive obstacles and avoid them autonomously. The existing UAV obstacle avoidance technology mainly includes infrared and laser ranging to realize UAV obstacle avoidance, but infrared and laser are easily affected and interfered by the external environment, resulting in inaccurate distance measurement and easy to cause safety accidents. In addition, the existing UAV obstacle avoidance technology generally relies on three-dimensional maps, binocular cameras or other high-pre...

AI Technical Summary

Problems solved by technology

[0009] In order to solve the above-mentioned problems in the prior art, that is, in order to solve the problems that the existing UAV obstacle avoidance technology has high cost, is not flexible enough, and the existing reinforcement learning method requires the control object to have strong fault tolerance, the present invention provides A multi-brain region collaborative autonomous decision-making method based on multimodal fusion, the method comprising:

Images