AI Utility Mapping for Buried Object Classification Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing utility locating systems struggle to efficiently process large sets of multifrequency electromagnetic data to accurately predict the type and source of buried objects due to limited on-board computing capabilities, requiring advanced AI processing that is currently unavailable.
Innovation Solution
Utilizing Deep Learning and Neural Networks with AI to analyze collected electromagnetic data, combining it with user-defined classifier data and ground truth data, to classify and predict the characteristics of buried utilities in real-time or post-processing, leveraging remote processing power through smartphones or cloud computing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If utility locating systems collect large sets of multifrequency electromagnetic data for accurate classification, then prediction accuracy is improved, but processing capability is insufficient due to limited on-board computing resources
Solution Approach 1:
The patent introduces a cloud-based processing system as an intermediary between the utility locating device and the AI processing requirements. The locating device collects electromagnetic data and transmits it to the cloud, where sophisticated AI models process and classify the data. This intermediary approach allows accurate prediction without requiring complex processing capabilities in the field device itself.
Solution Approach 2:
The patent divides the processing function into separate components: data collection occurs in the field using simple locating equipment, while data processing and classification are performed separately in the cloud. This segmentation allows the system to leverage powerful remote computing resources without burdening the portable locating device with complex processing requirements.
2Speed
If AI processing is performed locally in the utility locator, then processing speed is improved, but available processing power is insufficient for high-level AI applications
Solution Approach 1:
The cloud-based system serves as an intermediary that provides virtually unlimited processing power for AI model training and execution. The locating device simply needs to transmit collected data to the cloud, where sophisticated neural networks can process the data with the computational resources available in cloud data centers, eliminating the processing power constraints of portable devices.
3Measurement precision
If more processing power is available for AI analysis, then classification accuracy is improved, but device portability is reduced
Solution Approach 1:
By introducing a cloud-based processing intermediary, the system achieves high classification accuracy through powerful remote AI models while maintaining the portability of the field locating device. The cloud infrastructure absorbs the computational burden, allowing the portable device to focus solely on data collection and transmission functions.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enables accurate, real-time classification and prediction of buried utility types and ownership with high probability, overcoming processing limitations of traditional systems by harnessing the power of AI and remote computing resources.
Implementation Method 1
collecting utility and communication data by using utility locating equipment, or other electromagnetic receiving equipment, to gather and measure multifrequency electromagnetic signals from passive and active lines which are buried and/or underground
Data Source
AI summary
This disclosure relates generally to systems and methods for locating and mapping buried utility objects using Artificial Intelligence (AI). In an exemplary embodiment, electromagnetic data related to underground utilities and communication systems is collected and provided to a Deep Learning model to build a training set. New data is then collected and provided to the learning model to enable AI to make a prediction as to the type and location of any existing utilities. This predicted data may be used to create a map which may be stored and/or displayed. In some embodiments, the AI (Deep Learning model) processing may be performed locally in a Utility Locator and/or cable drum-reel, and/or remotely in a wireless device such as a mobile phone, laptop, vehicle, etc., and/or in the Cloud.