Technology Report 2022

Introducing Non-Intrusive Subsurface Mapping

Synopsis

This non-intrustive mapping technology deploys multiple sensors and AI – an approach akin to combining an MRI, a CT scan and an Ultrasound – to provide a 3D mapping of buried assets. Building a clear picture from complex data provides stakeholders the critical information they need to help prevent damages.

Exodigo’s non-intrusive subsurface mapping leverages multi-sensing, artificial intelligence (AI), satellites, drones and carts to create digitally geolocated 3D maps of underground facilities that can be integrated into any existing software, including GIS, AutoCAD or BIM (Building Information Modeling).
Exodigo’s AI analyzes signals detected from multiple sensors via satellites, drones and land carts to build a clear picture of subterranean environments, ultimately producing a geolocated file with multiple GNSS layers.
Non-instrusive subsurface mapping is anticipated to produce better, faster maps of underground environments at scale. It can be used to achieve Quality Level (QL) B in SUE projects, and reduce the potholing necessary to get to QL-A. To date, potholing has been reduced up to 50% on projects located in Israel, and reduced by 70% in Exodigo’s first project in the U.S. In addition, that method finds 20-50% more utilities (abandoned and live), as it removes the human bias and live interpretation in the field. The AI software finds all lines in the area, reducing issues caused by abandoned lines.

Back to Technology Report

Snapshot

Company Name(s): Exodigo, Inc.
Website(s): https://www.exodigo.com/
Technology Categories: Locating, Mapping, Positioning, Software
Stakeholder Target Audience: Designer, Facility Owner
Key terms /phrases: 3D map, abandoned line, AI, AutoCAD, BIM, GIS, GNSS, LIDAR, multi-sensor, pothole, SUE, unexpected utilities
Level of Production:
[technology is in full production and available for purchase]

Contact Case Study Author

Download Case Study PDF

Introduction Current Progress Potential Solution How the Process Works Results Pushing the Limits

Introduction

Exodigo is a startup and innovator combining advances in 3D imaging, sensors, and AI, to effectively map the underground, from man-made pipes and cables to soil layers, rocks, minerals, and even groundwater under any terrain. We deploy multi-sensor technology and AI. Our approach is akin to combining an MRI, a CT scan, and an Ultrasound, all at once, to produce one image which offers a clear view of the underground. In this case study we introduce the concept of Non-Intrusive Mapping for underground utilities, which is a combination of AI based advanced geophysics and satellite information. We then compare it with the current commonly available solutions.

Today’s current process

In this case study, we will compare the current process versus our Non-Intrusive Mapping method as tested at a real-life demo site in collaboration between Exodigo and the Israeli government.

Currently, there are a number of complex steps taken when mapping the underground, all of which have their limitations, and involve the potential for human error.

Test case: Burying a new line

Step 1: Assembling record drawings for all utilities in the area Subsurface Utility Engineering (SUE) Quality Level (QL) D[1] maps. This process involves contacting the utility and owners, receiving a record of their asset and building a single GIS file with all the layers.

Step 2: In this step someone goes out into the field to see if the maps are in line with what they see in person. This is SUE QL-C and is focused on getting visuals from the site. During this process, it sometimes reveals that step one was not accurate, and this creates doubt about what else is missing.

Step 3: Locators are called in, and they locate the lines, upgrading the maps to SUE QL-B. Their default is to use GPR and electromagnetics, open manholes, induce the lines, and follow them. A non-accurate 2-D image is the result, and oftentimes it’s just a confirmation of what was covered in steps 1 and 2, with nothing new discovered. This process can be time consuming and is sometimes dependent on professional interpretation in the field.

[1] See CGA Best Practice 2-14.

_{Quality Level D+C - Visual Confirmation: The sewer line (the line on the right side) is not where it is supposed to be (it’s 10 feet away). This is confirmed by the manhole located in the black circle.}
_{Quality Level B - The 6 short black lines are the lines located by the mapping team.}

Step 4: In this step the excavation itself occurs. “Quality Level A” is the last step, and excavation usually uses hydro-excavation or vacuum excavation. This is the way to verify the “Quality Level B” map and understand the depth of the utilities. The challenges with this method are the road closures involved, the impact on the environment and the fact it is akin to “looking through a straw.”

The challenge with this current process is that some issues with some of the buried utilities remain undiscovered. In the specific case highlighted it was a 45% miss rate. Five lines which were not on record were not located by the locator technicians, but were discovered in the case studies. These lines belonged to the city and there was no known record or maps of their existence.

In the black circle – lines missed by the locators

Why does this happen? We think it is because every step of the process depends on the previous one, where one mistake can set the whole thing off. Oftentimes, those going through the mapping process don’t see the full picture, and don’t know what they don’t know. This is a problem with abandoned lines. The main alternative today has been to trench the whole area, and dig the entire street, which can result in high costs and safety challenges. GPR might have a chance but has been known to have limitations and requires a level of interpretation skills. Locators used traditional GPR during the test case and missed the five utilities. It’s very hard to find what you don’t know exists.

Potential Solution

Non-Intrusive Mapping is an approach that changes the flow of the current process. Mapping the underground becomes “area-based,” as the technology requires no prior information or maps. The result is a digital geolocated 3D map of buried assets, which can be integrated into any existing software such as GIS, AutoCAD, or BIM (Building Information Modeling).

Exodigo’s ability to process and model large volumes of complex heterogeneous data allows us to decipher the underground world with accuracy. We are able to map out an area with no prior information utilizing tools such as advanced geophysics, satellite remote sensing and historical data, thermal imaging, and lidar. Advancements in artificial intelligence (AI) enable us to create one single map out of all the signals. We pair each sensor with a high accuracy GPS, in order to measure everything within 1 inch of precision so we can stack the signals into one GNSS map.

In order to measure the signals, the sensors have to be moved above the area of interest. Sensors can be moved either by satellites, airplanes, drones or land platforms.

The diagram shows all the physics affected by the underground. The more common ones for utility mapping are GPR (radar), and electromagnetics. Similar to modern GPR’s that can output a digital geolocated signal, Non-Intrusive mapping uses all the physics, or hybrid combinations, and measures a digital geolocated signal for every single one of them. Sensors are chosen for maximum sensitivity, and mainly come from the oil & gas, and mineral exploration industries.

Exodigo uses satellites in our initial analysis (3-5 feet max. accuracy), but true accuracy (5-20”), 3D and completeness are only achieved by using drones and carts. When it comes to underground infrastructure the difference between those distances can result in damages, so accuracy is key.

How the Process Works

We scan the area with all the sensors, using a drone, or a cart, and sometimes with both.

The data is then imported to the cloud, and our AI builds a clear picture from the gathered signals. We’re often asked how we train our AI, (as you need data sets). To do so we have used 100+ sites in Israel that were made available by government utilities and our strategic partners, and as such we received high accuracy as-builts and lidar scans of the buried assets to use as ground-truth. This is critical for developing an accurate AI model.

_{Scanning the area of interest in a crisscross pattern – with no assumptions on the number of utilities or their location}

Results

Each line has its own accuracy and certainty level which is outputted. All inaccuracies are taken into account and outputted in the results. Non-Intrusive Maps are Quality Level B+ (B1 & B2), as they use advanced geophysics, as recommended by the ASCE SUE guidebook.

The product is a geolocated file with multiple GNSS layers (with DGPS – 1” accuracy):

High accuracy image of the area, Lidar scan of the area, Dig\No dig layer, Utility location, Utility properties

The first layer is called Dig/No dig, and generates a colored-coded map, along with images of the pipes underground. In the green zone – the AI algorithms are certain there are no utilities (up to a certain depth, which is outputted by the algorithms). In the red zone – there is a high probability of utilities – no dig zone. The yellow zone is where the algorithms outputted a lower level of certainty, meaning it’s an area recommended for further investigation and potholing.

Exodigo’s Non-Intrusive Mapping solution discovered all the lines at the demo site (checked with potholes). The key is combining everything possible in order to create a single multi-sensing solution.

The next layer includes the depth of utilities (if successfully calculated by the AI) and physical properties: Conductive material, metallic material or none (plastic, etc.). Each depth is given its own accuracy. Accuracies range from 5” to 20”, depending on the terrain, congestion, background noise, and utility type. The AI calculates its accuracy for every single line and outputs it in the layers.

Since the file and layers are 3D, it’s possible to display it in 3D (along with the geolocated LIDAR scan) in AutoCAD, BIM, Google Earth, VR\AR software, or any other 3D visualization software.

Pushing the Limits

We are working on improving the speed of scan, speed of processing and accuracy of the solution.
Our goal is to build a product that is meaningful to the community and will potentially help reduce damages to infrastructure.

Left image – Congested areas training sites & output.
Right image – the 100,000 sq.ft. area mapped in a single day.

Another method to scan an area would be to use drones as a platform. Using the same exact sensors, vast terrains can be scanned at scale.

An area mapped with drones for an underground pipe and a power line

In conclusion, Non-Intrusive Subsurface Mapping will produce better, faster maps at scale. Potholing to achieve SUE QL-A can be reduced to the yellow zone. Potholing has been reduced up to 50% on projects in Israel, and a 70% reduction was achieved in the first project in the U.S.

Back to Technology Report