AI, a new partner in fiber deployment

2023-10-12 16:11:18

October 12, 2023 Today at 6:10 p.m.

Unifiber, which is responsible for the deployment of Proximus optical fiber in Wallonia, is now using AI on its construction sites to avoid costly repairs.

With its 600,000 homes to be connected to optical fiber, the logistical challenge for Unifiber, the Proximus joint venture, is far from simple. If leaders assure that the deployment is on the right track, any innovation is welcome to facilitate deployment.

Unifiber has just found a very useful one. They found it in artificial intelligence. It has just signed a partnership with Deepomatic, a Parisian company specializing in the integration of AI. She developed a construction site monitoring application for the Walloon company. Concretely, site managers are invited to take a photo every ten meters of the progress of the site at specific times.

Read also

Brice Le Blévennec installs artificial intelligence art in the living room

The AI, integrated into the app, has been trained on more than 5,000 site photos and is responsible for validating or not each step of the fiber installation. “In total, it analyzes eight different stages“, says Arnould Monard, works manager who has been testing the technology for several months. “We validate the depth of the trenches, the correct placement of the fiber without twisting or even the use of yellow sand”, lists the site manager.

Trained AI

In a few weeks, the system should be generalized to all teams. “The test phase lasted five months, because we had to have enough photos to train the AI. We have reached a sufficient level even if it will still have to be trained on certain aspects, including depth measurement trenches”, explains Arthur Journé, solution architect at Deepomatic.

Launched in 2014, the company has become a reference in the field. Its clients abroad include several leading operators, including Swisscom, Vodafone Spain and a large number of French operators.

“If a project has been poorly carried out, we only know it once it is completely finished, when it is ‘connected’ to the network.”

Nico Weymaere

CEO d’Unifiber

The main advantage of the validation system is that it avoids installation hiccups, which are costly for the company. “If a project has been poorly carried out, we only know it once it is completely finished, when it is ‘connected’ to the network,” says Nico Weymaere, CEO of Unifiber. “Finding the origin of the problem often requires major work and the reopening of trenches.. This necessarily has a cost and a frustrating side for the municipality and local residents.”

In addition to ensuring that each step is carried out correctly, the process allows Unifiber to have the guarantees proving that the work was carried out correctly by its teams. “The workers do not see the use of the app as a constant check. I was already taking photos which I stored in my cloud. They were useful to me in the event of a problem to prove that the work was carried out well “, explains the works manager.

More construction sites

By extending the use of AI to all teams, much more detailed monitoring of construction sites will allow Unifiber to better monitor expected growth. The group plans to move up a gear in the coming months. “This self-control will therefore make it easier for us to spread ourselves across more different sites at the same time,” explains the manager.

As all of the data collected is geolocated, it will also be valuable during the following stages. “When a customer takes out a fiber contract and it is necessary to make the last connection to their home, we will know exactly the state of the network in front of my homei”, we specify at Unifiber.

Within five years, the group hopes to have reached 250,000 homes connected to fiber. Currently present in 18 municipalities, Unifiber plans projects in 30 municipalities by 2023 and 45 by mid-2024.

1697133738
#partner #fiber #deployment

Share:

Facebook
Twitter
Pinterest
LinkedIn

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.