Visual positioning simulation method and system based on deep learning in irradiation environment

Abstract

The invention discloses a visual positioning simulation method and system based on deep learning in an irradiation environment. The system comprises a nuclear heat chamber support platform, an industrial binocular camera and a PC. The industrial camera acquires a video stream of the operation platform, a PC extracts pictures frame by frame, position information of different types of target objects is acquired by adopting a visual algorithm, and reconstruction is performed in a VR platform based on a Unity 3D engine. The method comprises the steps of performing calibration and distortion correction of a binocular camera; eliminating image salt and pepper noise by a median filtering preprocessing algorithm of the polluted image in a nuclear environment; adopting a yo.v4 target detection and BM binocular stereo matching fusion algorithm to generate a target type and a three-dimensional position. According to the method, a yolo.v4 deep learning algorithm is adopted, the detection speed exceeds that of a traditional visual detection algorithm, the types of the identified objects can reach infinity theoretically, multi-target identification, positioning and VR real-time reconstruction of the nuclear heat chamber are achieved simply through visual information, and the hot chamber manipulator is guided to work on the target objects.

Description

technical field [0001] The invention belongs to the field of nuclear industry intelligent simulation and real-time image processing, in particular to a visual positioning simulation method and system based on deep learning in an irradiation environment. Background technique [0002] As a major equipment for remote operation of nuclear power maintenance, nuclear power manipulators can be used to perform heavy and complex work in extremely harsh and strong radiation environments where humans cannot (or are restricted) to enter, especially for spent fuel reprocessing and decommissioning of nuclear facilities. [0003] Although the manipulator has experienced more than 50 years of development since its appearance, many different types of manipulators have been born. They have shown their talents in various working environments and already have a mature and complete technical system. However, the research on nuclear industry power manipulator in our country is still in its infanc...

AI Technical Summary

Problems solved by technology

However, my country's current research on nuclear industry power manipulators is still in its infancy.

The reason is to a large extent: the operation of large-scale manipulators, especially in high-risk nuclear environments, in order to ensure safety and precision, it is necessary to obtain sufficient information about the environment and target objects to guide the operation of the manipulator, which only relies on the operator's naked eyes Manual operation with experience is not enough; even if there are some relatively mature remote operation simulation control systems for manipulators in the market

However, due to the action of high-energy particles in the nuclear environment, most of the precision sensors that are essential for environmental data collection will greatly reduce their lifespan or even fail, so simple industrial cameras can only be used to collect such data

At this time, if we simply use traditional machine vision image processing technologies such as OpenCV edge detection and other related algorithms to obtain relevant information on the collected data, the results will be poor, low in precision, and slow in speed, requiring a large number of feature extractions. Segmenting a single target object is also subject to conditions such as illumination, distance, etc. Obviously, it is far from enough to meet the information acquisition needs of the entire target environment and various types of operating target objects

Images