A project that is embracing AI technology to improve safety at tramway crossings is to move forward to the next stage of development as part of a partnership between the Light Rail Safety and Standards Board (LRSSB) and Starling Technologies.
The project will see the installation of advanced monitoring cameras at key locations on several tram networks. These will then be used to observe and analyse the behaviour of pedestrians at Non-Motorised User (NMU) crossings to provide insights into how they’re used, identifying potential risks and opportunities for safety improvements.
Craig O’Brien, Head of Engineering Safety and Innovation at LRSSB, said: “This project is a significant step forward in our mission to improve safety across the light rail sector. Understanding user behaviour at NMU crossings will help us develop data-driven strategies to reduce risks and ensure safe interactions between light rail vehicles and pedestrians.
“The findings from this project will contribute to a broader discussion on NMU safety, potentially shaping future policies and technologies to enhance light rail operations nationwide.”
At the moment the locations of the cameras are not being revealed at this would risk influencing the behaviour of people using the crossings. However, this has no additional impact on privacy as the crossings are already monitored for safety purposes.
The collected data as part of this project will help to inform the sector on best practices for managing NMU safety in the future.
The state-of-the-art camera systems and analytical tools are being supplied by Starling Technologies, a leader in smart surveillance solutions. Andrew Caleya Chetty, Founder and CEO of Starling Technologies, said: “We are really excited to be helping these next-generation thoughts and actions, to help keep people safer around tramways. Working with LRSSB and tram operators, we are convinced we will identify areas of improvement.”
Like so many things labelled as “AI”, this is NOT true artificial intelligence. It does not autonomously absorb feedback, to learn and adapt its actions accordingly in real time. There may well be some high level data-processing and machine analytics secondary to the data capture, but it is implementing pre-programmed frameworks, not developing it’s own.