Channel monitoring method, electronic device, and storage medium

Patent metadata

Field	Value
Publication number	US-12596383-B2
Application number	US-202418815750-A
Country	US
Kind code	B2
Filing date	Aug 26, 2024
Priority date	Jul 6, 2023
Publication date	Apr 7, 2026
Grant date	Apr 7, 2026

Abstract

Provided is a channel monitoring method, an electronic device, and a storage medium. The method includes: obtaining scan data collected by a vehicle body collection component of an automated guided vehicle (AGV) in an area around a vehicle body; obtaining video data collected by a camera in a video collection area, where the camera is one of a plurality of cameras and is used to collect video of at least a partial area of the channel to be monitored of the plurality of channels to be monitored; in a case of determining an existence of a target object based on the scan data collected by the vehicle body collection component and/or the video data collected by the camera, obtaining a target channel where the target object is located; and generating target warning information for the target channel where the target object is located.

First claim

What is claimed is:

1 . A channel monitoring method, comprising:
obtaining scan data collected by a vehicle body collection component of an automated guided vehicle (AGV) in an area around a vehicle body, in a case of determining existence of a first target object in the scan data based on the scan data, obtaining a first target channel where the first target object is located,
wherein the AGV is one of a plurality of AGVs, the AGV of the plurality of AGVs travels in any one channel to be monitored of a plurality of channels to be monitored, the plurality of channels to be monitored are a plurality of channels required to transport a yarn spindle in a yarn spindle production workshop, the area around the vehicle body includes at least a partial area of the channel to be monitored in which the AGV is traveling, and the first target channel is one of the plurality of channels to be monitored;
obtaining video data collected by a camera in a video collection area, in a case of determining existence of a second target object in the video data based on the video data, obtaining a second target channel where the second target object is located,
wherein the camera is one of a plurality of cameras and is used to collect video of at least a partial area of the channel to be monitored of the plurality of channels to be monitored, the video collection area includes at least a partial area of the channel to be monitored which is collected by the camera, and the second target channel is one of the plurality of channels to be monitored;
determining whether the first target channel where the first target body is located is a same as the second target channel where the second target body is located; in a case of the first target channel is the same as the second target channel, generating hybrid warning information for the first target channel and the second target channel; in a case of the first target channel is different from the second target channel, for the scan data, generating scan warning information for the first target channel, and for the video data, generating video warning information for the second target channel, wherein priorities of the hybrid warning information, the scan warning information and the video warning information gradually decreases.

2 . The method of claim 1 , further comprising:
inputting a scan image included in the scan data collected by the vehicle body collection component into a first detection model to obtain a first detection result, wherein the first detection model is used to identify whether there is at least a part of partial features of a target object in the scan image, and the first detection result represents whether the first target object exists in the scan image included in the scan data.

3 . The method of claim 2 , wherein inputting the scan image included in the scan data collected by the vehicle body collection component into the first detection model to obtain the first detection result, comprises:
preprocessing the scan image included in the scan data collected by the vehicle body collection component to remove an edge area in the scan image and obtain a sub-image corresponding to a center area in the scan image; and inputting the sub-image corresponding to the center area in the scan image to the first detection model to obtain the first detection result, wherein the first detection model is used to extract features from the sub-image corresponding to the center area in the scan image, obtain a feature map of the center area, and detect whether there is at least a part of partial features of the target object in the feature map of the center area.

4 . The method of claim 3 , wherein the first detection model comprises at least a first network layer, a second network layer, and a third network layer,
wherein the first network layer is used to perform feature extraction on the sub-image corresponding to the center area to obtain at least one feature map corresponding to the sub-image, wherein a quantity of the at least one feature map is negatively correlated to a size of a feature block of the at least one feature map, the second network layer is used to predict whether there is at least a part of partial features of the target object based on the at least one feature map corresponding to the sub-image, and obtain a probability value, and the third network layer is used to process the probability value to obtain a first value or a second value, wherein the first detection result is the first value or the second value, the first value is used to represent the presence of at least a part of the partial features of the target object, and the second value is used to represent the absence of at least a part of the partial features of the target object.

5 . The method of claim 2 , further comprising:
inputting the video data collected by the camera into a second detection model to obtain a second detection result, wherein the second detection model is used to identify whether the target object exists in the video data, the second detection result represents whether the second target object exists in the video data, and a time complexity of the second detection model is greater than a time complexity of the first detection model.

6 . The method of claim 5 , wherein inputting the video data collected by the camera into the second detection model to obtain the second detection result, comprises:
preprocessing the video data collected by the camera to obtain a video image set, wherein the video image set includes at least one group of two adjacent frames of video images, and the two adjacent frames of video images meet a requirement that a difference value between feature maps of the two adjacent frames of video images is greater than a preset difference value; inputting each video image in the video image set to the second detection model to obtain a sub-result corresponding to the video image; and in a case of at least one sub-result among sub-results of all video images in the video image set is a third value, determining the third value as the second detection result; or in a case of a sub-result of each video image in the video image set is a fourth value, determining the fourth value as the second detection result, wherein the third value is used to represent the existence of the target object, and the fourth value is used to represent the existence of no target object.

7 . The method of claim 6 , wherein the second detection model comprises at least a fourth network layer, a fifth network layer and a sixth network layer,
the fourth network layer comprises at least two dense layers connected in sequence, wherein a connection feature map of a previous dense layer is used as an initial feature map of a next dense layer, the at least two dense layers are used to perform convolution processing on the initial feature map to obtain a convolution feature map and perform connection processing on the initial feature map and the convolution feature map to obtain a connection feature map, and the initial feature map of a first dense layer of the at least two dense layers is obtained based on the video image, the fifth network layer is used to perform feature aggregation on the connection feature map of a last dense layer of the at least two dense layers to obtain a target feature map, and the sixth network layer is used to predict the target feature map to obtain the third value or the fourth value.

8 . The method of claim 1 , after obtaining the second target channel where the second target object is located, further comprising:
determining whether there is an AGV traveling in the second target channel, wherein for the video data, generating the video warning information for the second target

Assignees

• Zhejiang Hengyi Petrochemical Co Ltd

Inventors

• Wang Peng
• Peng Xiantao
• Li Dake
• Liu Mingyi
• Xu Feng
• Shen Jun

Classifications

• G06T2207/20084

  Artificial neural networks [ANN] · CPC title

• G06T2207/20021

  Dividing image into blocks, subimages or windows · CPC title

• G06T2207/10016

  Video; Image sequence · CPC title

• G05D2111/10

  Optical signals · CPC title

• G05D2105/20

  for transportation · CPC title