🧠 Google’s AI Climate-Proofs Your Commute

Subscribe | Contact | Meet The Team

Welcome back AI prodigies!

In today’s Sunday Special:

• 🚘Innovative Intersections

• 📍Local Limitations

• 🚥Beyond Green Light

Read Time: 6 minutes

🎓Key Terms

• Supervised Machine Learning: an algorithm that predicts the future based on exactly what happened in the past by monitoring the relationship between inputs (i.e., causes) and outputs (i.e., effects).

• Internet of Things (IoT): a network of sensors and interrelated devices that connects, collects, and exchanges data stored in the cloud.

• Hydrologic and Inundation Models: flood forecasting models that predict the amount of water flowing in a river, which geographic areas will be affected, and how deep the water will be.

🚘INNOVATIVE INTERSECTIONS

The average American driver emits 4.6 metric tons of carbon dioxide each year. Suppose you’re among the 4% who drive Electric Vehicles (EVs). Congratulations! For us ordinary folks, however, emitting an elephant’s worth of carbon dioxide may invoke anguish, concern, or apathy. To compound matters, widespread EV adoption in the U.S. will likely not occur until at least 2050, though probably much later. Luckily, for worried readers, there’s a hassle-free solution to reduce your intersection stops by 30% and your emissions by 10%; separately, peruse these 10 Google product enhancements to up your sustainability game. To the apathetic, Google’s clever AI-driven solutions promise to reduce the time you spend waiting at intersections.

Meet Project Green Light. It retimes traffic lights by applying supervised machine learning to Google Maps data, so you can spend less time in traffic, emitting less carbon dioxide. Green Light is available at 70 intersections worldwide, aiming to dampen emissions from road transportation, which causes nearly 25% of greenhouse gas emissions in the U.S. and 15% globally. Though the proportion of vehicle emissions from intersections is unclear, pollution at city intersections can be 29 times higher than on open roads, and accelerating after stopping accounts for half of intersection emissions. Testing data from 12 global cities—Abu Dhabi, Bali, Bangalore, Budapest, Haifa, Hamburg, Hyderabad, Jakarta, Kolkata, Manchester, Rio de Janeiro, and Seattle—indicates a potential 30% reduction in stops and 10% reduction in emissions after adopting Green Light’s recommendations.

Some recommendations implore traffic engineers to add two seconds between the start of one green light and the next, allowing more vehicles to pass through both intersections without stopping. Others involve multiple steps, altering the frequency and duration of several traffic lights in a high-volume corridor. Green Light’s system enables monthly changes as traffic patterns shift since real-time adjustments require more conclusive testing data, and most traffic infrastructures don’t support it. City engineers validate suggestions by using traffic counts from video footage or Internet of Things (IoT) sensors.

📍LOCAL LIMITATIONS

When building traffic timing recommendation systems, researchers make tradeoffs between customization and accessibility. Unfortunately, Google hasn’t let the public peek under the hood of their AI models. However, we know they programmatically build a model of each intersection, including its structural layout (e.g., lane volume and type or constraints like U-turn restrictions) and the interactions between traffic patterns and light scheduling. Then, they integrate all of a city’s models into an interactive system so Green Light can analyze them in tandem in real time. Beyond city-based customization, local preferences, which are impossible to capture through Google Maps driving data, often dictate traffic flows.

Although traffic engineers can implement free suggestions for each intersection in as little as 5 minutes, some driving environments present acute challenges. Some recommendations in Manchester, United Kingdom, are unhelpful because they fail to account for the city’s prioritization of buses, cyclists, and pedestrians across the busiest of its 2,400 intersections. Though some filters try to block suggestions that deprioritize pedestrians, they’re inadequate. To incorporate local preferences into the recommendation system, Google must create a generic model and distribute it to each city for customization. However, most cities lack adequate talent—traffic engineers with machine-learning abilities—to add parameters to the model and fine-tune it over time. Nevertheless, air quality at less busy intersections rose by 18%.

On the other hand, in Kolkata, India, traffic police implemented 13 Green Light recommendations and received positive reception from motorists. Human traffic coordinators dictate traffic flow on the chaotic streets of India’s megacities, with the Indian traffic controller market size expected to triple by 2029. Variables that predict CO2 emissions at intersections, namely the frequency of stops and the timing of stop-and-go cycles, are far from optimal. Furthermore, road transport accounts for 20 to 30% of India’s urban air pollution. Google’s researchers expect Green Light to reduce emissions at higher rates than the U.S., which stands at 10%.

🚥BEYOND GREEN LIGHT

Despite the affordability and scalability of Green Light’s solution, emissions from road intersections remain a drop in the bucket of global greenhouse gas emissions. Fortunately, Google’s incomprehensibly vast datasets and strategic deployment of AI have seeded several initiatives to address climate-related challenges. Here are the top two:

1. Fuel-Efficient Routing: Route inefficiency plagues both ground and air transportation. Drivers take hilly, traffic-riddled routes, leading to excess acceleration. In Google Maps, fuel-efficient routing suggests more environmentally friendly ways to arrive at destinations, preventing CO2 emissions equivalent to taking 500,000 ICE vehicles off the road. In the air, pilots leave contrails—thin, white lines of airplane exhaust—which account for over a third of aviation emissions. In conjunction with American Airlines, Google Research used machine learning to develop contrail forecast maps for pilots, halving the number of contrails produced.

2. Predicting Floods: Despite receiving just 9% of U.S. news coverage of disasters, floods are the most common natural disaster. Leveraging satellite imagery, weather forecasts, and the Hydrologic and Inundation Models, Flood Hub provides 7-day river flooding forecasts to over 460 million people across 80 countries. With England facing the most dangerous floods in over a decade, Flood Hub predicts when and by how much the Thames River will rise.

🩺 PULSE CHECK

📒FINAL NOTE

If you found this useful, follow us on Twitter or provide honest feedback below. It helps us improve our content.

How was today’s newsletter?

• ⭐️⭐️⭐️⭐️⭐️Nailed it!

• ⭐️⭐️⭐️Average

• ⭐️Epic fail.

❤️AI Pulse Review of The Week

🎁NOTION TEMPLATES

🚨Subscribe to our newsletter for free and receive these powerful Notion templates:

• ⚙️150 ChatGPT prompts for Copywriting

• ⚙️325 ChatGPT prompts for Email Marketing

• 📆Simple Project Management Board

• ⏱Time Tracker