Recognition method for environment self-adaptive navigation scene of miniature unmanned aerial vehicle based on function intelligence

Abstract

The invention discloses a recognition method for an environment self-adaptive navigation scene of a miniature unmanned aerial vehicle based on function intelligence. A combined flight platform composed of the miniature unmanned aerial vehicle and an intelligent mobile terminal is designed, a non-visual sensor of the intelligent mobile terminal is used for collecting data and extracting environmental features from the data, the environmental features are accurately recognized through a trained SceneNet deep learning network, and a visual sensor of the intelligent mobile terminal can also be used for collecting images. Collected images are extracted and classified through a trained mobileetv2 network, so that a final recognition decision can be obtained. Compared with a traditional scene recognition method, a double-layer scene recognition method has the advantages that the accuracy of scene recognition can be improved by simulating the physiological behavior perception environment of living beings, functional intelligent navigation with good adaptive performance in different environments is achieved, adaptive navigation software can be freely developed on the premise that a navigation system is not changed, and the application prospect is wide. Effects of compact flight and reliable operation in different environments are achieved.

Description

technical field [0001] The present invention relates to the technical field of navigation scene recognition, in particular to a method for environment adaptive navigation scene recognition of a micro UAV based on functional intelligence. Background technique [0002] With its superior flexibility and hovering ability, micro UAVs are widely used in surveillance, search and rescue missions in various environments. In order to perform these tasks with high precision, micro-drones must have the ability to fly autonomously. Inertial sensors are autonomous sensors whose measurement information is not affected by the external environment. Fully considering the cost and payload constraints, most micro-UAV navigation systems are equipped with low-cost, low-precision MEMS inertial sensors. In order to improve the accuracy of inertial navigation, most indoors adopt the visual / inertial integrated navigation method, and most outdoors adopt the GPS / inertial integrated navigation method....

AI Technical Summary

Problems solved by technology

In low dynamic applications, feature extraction for scene recognition based on deep neural network is significantly better than the current state-of-the-art methods, but when the environment changes drastically under high dynamics, scene recognition may need to be redesigned

Images