Inspiration

Traffic is shitty -- just for example, let's take New York. With the highest population and the second-worst per-person hours spent in traffic, New York City earns the dubious honor of losing the most total hours to congestion of any city in the United States. The average New Yorker spends more than 23 minutes a day idling, costing the city 52,000 jobs and 20 billion dollars while emitting 15 million metric tons of carbon dioxide per year. Preliminary studies found that traffic signal optimization could save up to 25% of all transit time, and more than 99 percent of traffic lights work on preset timer systems that do not adapt to real-time traffic conditions. I’m sure you’ve been there, getting stopped by a red light even though you’re the only car on the road.

What it does

MAXTHRU optimizes traffic signals in real time using data from CCTV cameras. Cloud API counts the number of vehicles waiting in each lane; then that data is used to generate signal timings that maximize throughput and minimize idling time.

How we built it

We used Google Cloud API to demonstrate car detection and counting, and Unity with C# to demonstrate the use case and simulated throughput of a single intersection before and after the implementation of MAXTHRU.

Challenges we ran into

As we're all super beginner freshmen hackers, the whole experience was new and exciting to us! We spent a lot of time pivoting between different ideas (supply chain management? empowering refugee education? self-care centric social media?), but when we settled on this one we struggled with implementing the API and building a demo-able product. We also tried to use complex, pre-built city flow simulators like A/B Street and Cities:Skylines Traffic Manager: Presidential Edition, but it was difficult to navigate their existing codebases and implement personalized algorithms for signal timing. Connecting the CV and the Unity traffic simulation is something we are still working on implementing.

Accomplishments that we're proud of

We thought of and demonstrated the viability of a relatively simple, cheap-to-implement way to save billions of dollars, person-hours, and impatient carpoolers.

What we learned

We learned to use an API for the first time, use Unity with C# to make demos, how to (largely unsuccessfully) wade through Rust code for an existing simulator, and that the real hackathon was the friends we made along the way!

What's next for MAXTHRU

We are currently in contact with the head developer of the A/B Street traffic simulator to implement the MAXTHRU in the simulator, which will give us concrete data on how MAXTHRU can decrease average trip time and idle time in a sophisticated simulator.

Built With

Share this project:

Updates