Google is working with AI to digitise cable networks

Electricity is the energy that drives modern society but trying to figure out where cables are located continues to provide engineers with major challenges when considering that these cables were installed decades ago. Now, Google is using AI to automatically generate maps of where all cables are so that future engineers know precisely how to make upgrades and repairs. Why are cable networks providing engineers with significant challenges, what will Google’s new project do, and should this idea of digitised cables be brought into homes?

Why are cable networks providing engineers with major challenges?

If there is one fact of life that I am truly grateful is that electricity comes out of my sockets whenever I require it. Just like the Armstrong and Miller Time Traveller sketch, the number of people in society who are entirely oblivious to the effort and resources needed to correctly operate a nationwide grid is astonishing (electricity is that thing that comes out of the wall). However, this sheer complexity makes it difficult for those not involved with electronics to truly understand and appreciate.

Additionally, I am also grateful that network operators are able to perform repairs on cables regardless of the weather. Power may be lost during the middle of a storm, but most will find the power is restored within a day after engineers diagnose the issue, locate damaged cables, and then replace cables.

Trying to find a broken cable is a piece of cake when the damage has been physically spotted and reported by a member of the public, but what about if the cable is in a rural area, buried underground, or installed by a third party? It is in these instances that distribution network operators struggle as finding a damaged cable can be challenging.

Of course, equipment exists that can send pulses down a cable, measure the reflection, and then determine the distance of the broken portion from the tester. Other testers use electric field detection that operates similarly to metal detectors and can be used to pinpoint a break. But both methods will require detailed maps of where cables have been installed, and most cables in use were installed at least a decade ago. In fact, some cable network operators still rely on hand-drawn maps that are electronically scanned. Thus, finding damaged cables requires administrative staff to find the appropriate drawings and then send those off to a field engineer.

This problem is amplified for those looking to make upgrades or changes to a cable network. One typical example is the introduction of electric vehicle charge points; the installation of a new charging point requires detailed drawings of cables that are up to date.

Google developing AI with UK Power Networks to create a digital cable map

Recognising the challenges cable network operators face, Google recently announced a new project that uses their DeepMind AI to scan drawings and automatically create power line network maps that are fully digital and recognise individual cable installations. The company partnering with Google, UK Power Networks, is reported to have over 180,000km of cable currently mapped on drawings that are then electronically stored.

However, the electronic version of these maps is not interactive and does not separate the cable data from the map (essentially, a document scan). This means that while engineers can quickly obtain cable drawings, they are unable to separate the drawing into individual layers.

The new AI being developed by Google is scanning each document to automatically recognise where cable installations are, create digital maps of the installations, and provide metadata that allows for the separation of key details. Furthermore, the maps generated by Google will be provided to the public for free so that engineers needing to make upgrades (such as installing an EV charging point) can do so safely while also allowing utility workers to identify areas of potential danger.

According to Google, previous AI solutions have been unable to recognise hand-drawn straight lines representing cables, but their solution is able to do just this. Additionally, Google stated that their new system significantly reduces the number of hours needed to digitise maps from 20,000 to just 15 minutes.

Should infrastructure maps be brought into homes?

Using AI to automatically read cable drawings to create interactive digitised maps provides exciting opportunities for those involved with power distribution, but the use of digitising infrastructure in homes could provide homeowners with major benefits.

Cables and pipes buried in walls present homeowners with serious risk if not properly located. Driving nails or screws into a wall (whether it’s putting up a shelf or portrait) can penetrate cables causing electric shock or damaging pipes, causing internal leaking of walls. Under normal circumstances, this shouldn’t be an issue as building regulations require such infrastructure to be located at the edges of walls and in straight lines from sockets. However, rouge builders and DIY enthusiasts may not be aware of such regulations, and this often sees rouge cables and pipes located in areas they should not be.

As such, having a digital map of all infrastructure in a home would improve safety for anyone looking to make changes and provide critical data for those needing to perform repair work. Additionally, such a map could be combined with augmented reality that shows the wearer where services are located.

Overall, Google’s DeepMind demonstrates the continuing power of AI and how it can be used to convert ageing data formats into fully digitised variations with metadata allowing for better data analysis. But could this AI be expanded to also map user homes? Can digital twins of homes help homeowners make better decisions regarding their property?