Vladimir Krylov: AI Solution for Mapping Street Assets

12 January 2021
Vladimir Krylov: AI Solution for Mapping Street Assets

Dublin, 29 July 2020: Imagine you’re working for a utility company in a sparsely populated part of Donegal. The company has asked you to drive down a very lengthy stretch of country road and map out every single electrical pole. That’s going to take a while! Even when you finish, chances are you missed a couple poles. You’re only human, after all.

This is exactly the type of scenario that AIMapIT was designed to assist with. Rather than having an employee spend valuable time on the road, the AI system can be trained by the ADAPT research team to simply use available road images to identify and map street furniture, like road signs or telephone poles, and tag them with the precise GPS coordinates. This minimises the amount of human errors and makes the process much more efficient. The programme can be trained to recognise any number of assets, depending on the individual project’s needs and, in a world full to the brim of accessible information, the details can be what make or break you.

Last week, ADAPT Radio host Donal Scannell interviewed Vladimir Krylov, a member of the AIMapIT research team, and learned all about it. The episode is available on our SoundCloud channel or wherever you get your podcasts.

AIMapIT is an AI solution for discovering, detecting, and GPS mapping stationary objects. This is very important for companies that manage infrastructure assets, such as telecom operators, utility companies, and state agencies, as it will provide them with the exact location of all of their assets, even with companies whose infrastructure is distributed along roads and city streets. The collection of geographic information surrounding stationary street assets, involves searching through very large amounts of street level imagery, taking note of where there are assets and recording it.

This process done manually takes a huge amount of time and due to human error can be sometimes imprecise. The amount of time required to do this task also means that many companies often outsource this job to countries with cheaper labor, allowing them to hire more people and get more work from them for less money. The hours and personnel required seem really high when in just over a day AIMapIT managed to explore “over 10 thousand kilometres of rural Ireland”. Once it was fully operational, the programme had the potential to explore the whole of Ireland within just a few weeks.

AIMapIT is an ADAPT Centre spin-in project within Trinity College Dublin and funded by the Enterprise Ireland Commercialisation Fund. It is led by Trinity Professor in Statistics, Dr Rozenn Dahyot, managed by Julie Connelly and Moataz Ahmed, and made up of the following team members: Dr Vladimir Krylov, Conor McNally, Declan McKibben, Kieran Flynn, Dr Eamonn Kenny, and Dr Jing Su.

Share this article: