Target Position Management Using Visual SLAM and Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for managing the position of targets at construction sites require IC tags attached to each device, limiting the ability to track and manage devices without tags.
Innovation Solution
A system using a mobile device with an image capture unit attached to a helmet, which captures images and uses visual simultaneous localization and mapping (v-SLAM) and convolutional neural networks for target identification and position tracking, allowing for the management of targets without the need for IC tags.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If IC tags are attached to each device for tracking, then the position management of devices becomes reliable, but the system cannot manage devices without IC tags
Solution Approach 1:
The patent uses image capture devices to create visual copies of devices in the form of images. These images are then processed through neural networks to extract identifying features and position information, effectively creating a digital representation that replaces the need for physical IC tags on each device.
Solution Approach 2:
The patent replaces the mechanical IC tag reading system with an optical-based image capture and processing system. Instead of using physical tags and readers, the system uses cameras to capture images and neural networks to process them, thereby substituting a mechanical/electronic system with an optical-computational one that has greater versatility.
2Extent of automation
If IC tag readers are installed in elevators to track device movement, then movement tracking becomes automated, but devices without IC tags cannot be tracked
Solution Approach 1:
The patent creates a universal tracking system using image capture devices that can identify and track any device within the field of view, regardless of whether it has an IC tag. The neural network is trained to recognize diverse device types and features, making the system multi-functional and adaptable to various device configurations.
Solution Approach 2:
The patent introduces image processing and neural network analysis as intermediary steps between the image capture device and the position management system. This intermediary layer extracts meaningful information from images, enabling automated tracking without requiring direct interaction with IC tags on each device.
3Adaptability or versatility
If image capture devices and neural networks are used for target identification, then the ability to manage targets without IC tags is improved, but the system complexity increases
Solution Approach 1:
The patent implements a self-service system where the image capture device automatically captures images, the neural network automatically processes and identifies devices in the images, and the system automatically updates position information. This automation reduces the need for manual intervention and simplifies operation despite the underlying computational complexity.
Solution Approach 2:
The patent performs preliminary actions by pre-training the neural network with large datasets of device images and features before deployment. This preliminary training enables the system to quickly and accurately identify various devices without requiring complex real-time processing during actual operation, thereby reducing operational complexity.
Data Source
AI summary
A system for managing a position of a target stores identification information for identifying a target to be managed in association with position information indicating a position of the target. The system further obtains an image from an image capture device attached to a mobile device and obtains the image captured by the image capture device at an image capture position and image capture position information indicating the image capture position. The system further locates the position of the target included in the image using the image capture position information. The system further stores the position of the target in association with the identification information of the target.