[RahulNagaraj]

According to a media report, Google is using AI algorithms to utilise traffic-related data from 'Google Maps' to regulate traffic flow in major cities across the world.

https://www.youtube.com/watch?v=T6c_NdpbUvE

The new AI-enabled traffic lights are part of 'The Google Green Light Project', and are already active in 12 cities worldwide, including Seattle (USA), Rio de Janeiro (Brazil), Haifa (Israel), Abu Dhabi (UAE), Bali & Jakarta (Indonesia), Bangalore, Hyderabad & Kolkata (India), Manchester (UK), Hamburg (Germany) and Budapest (Hungary). The project is said to be affecting 30 million car journeys per month.

As per reports, using Google Maps, the AI takes into account the structure, traffic patterns and light scheduling of each intersection to come up with recommendations for optimising the traffic lights. Based on these recommendations, a local engineer can then tweak the timing of the red light based on the flow of cars. The tweaks can even be coordinated with several adjacent intersections to ease traffic congestion.

The Green Light Project states that the AI-based model doesn't require any updates on the existing infrastructure and can be implemented at no extra cost. This is a significant improvement considering previous methods of optimisation usually included a bunch of sensors or manual vehicle counts.

Google suggests that as per early numbers, the AI-based traffic lights have the potential to reduce stops at intersections by 30% while cutting down emissions in these areas by 10%.

The tech giant wants to expand the project to more cities in 2024. Google is asking city traffic engineers around the world to fill in a form and enter the waiting list.

Apart from AI-based traffic lights, Google has also announced a few other transportation-related initiatives. The company stated that its 'fuel-efficient routing' on Google Maps will be expanded to India & Indonesia this year. Recently, Google also released a new tool which offers more insight for EV shoppers.

Source: CarScoops

Link to Team-BHP news

[shancz]

Now this is an area where AI can shine and start delivering immediate results across traffic geographies.

I don't see any reason why it can't be truly autonomous once it has sufficient data and adapt itself to the deployment environment. The possibilities are immense in solving real world problems

[nicodash]

Originally Posted by shancz

I don't see any reason why it can't be truly autonomous once it has sufficient data and adapt itself to the deployment environment.

I suspect it's not truly autonomous because it would require the installation of hardware/software/devices that can receive instructions from Google's algorithm and set the red/green light timings in the existing signal management software. Given the various types of hardware/software systems worldwide for managing signals, it might not be possible for Google to support integration with all of them. These existing systems may also lack the capability to take inputs from a third-party source (i.e., Google) and adjust signal timings accordingly.

The better alternative would be for Google to provide its own hardware-software stack for signal management. This would be a more technologically sound solution, but it could raise privacy concerns globally and necessitate complex negotiations and tender processes in each jurisdiction. In a country like India, this would involve working with thousands of municipalities/councils and potentially bidding on tenders.

However, if Google can implement either of these solutions to automate the process, I believe that traffic congestion and emissions might decrease more than what they claim with the manual approach, as the AI could dynamically adjust timings based on real-time traffic patterns, not just relying on past traffic history.