Roundabout Technologies

tl;dr: Roundabout Technologies makes AI-powered traffic signal control systems that optimize lights based on current traffic conditions in real time.

We’re Collin and Sabeek, the co-founders of Roundabout Technologies. We have a combined 15 years of industry experience in embedded systems and computer vision, building safety-critical software that’s been deployed on medical devices at Verily and in driverless vehicles at Waymo and Zoox. As avid cyclists, we’ve long taken an interest in urbanism and public infrastructure, and we both have a deep desire to see cities become more pedestrian and bike-friendly. Roundabout sits at the intersection of our interests and our professional skills, a chance for us to use our hardware and AI skills to make cities more livable.

The Problem

Cities want traffic infrastructure that works for their residents. But existing systems fall short. Drivers waste hours at inefficiently programmed traffic lights. This floods city governments with resident complaints, and can even cause safety issues as impatient drivers race to catch yellow lights and risk causing crashes in the process. Existing light timings have to come from traffic studies that measure traffic flows for a week and use that to determine light timings for the next 5 years, a workflow that is both expensive and imprecise.

While induction loops and existing camera-based systems make things somewhat better, they are still brittle and manually programmed, so they scale poorly on complex, highly congested areas. These detection systems have limited range, unable to see cars before they’ve already come to a stop at an empty red light. Pedestrians and bikes are rarely detected in these systems, so cities must add extra buttons as an afterthought to respond to pedestrian traffic. And providing priority for emergency vehicles and buses requires extra transponders be installed.

Add all this up, and you get a system that requires lots of expensive hardware and manual programming but still leaves drivers, pedestrians, and transit riders deeply unsatisfied. Clearly we should be able to do better than this. And now, we can!

Solution

Roundabout Technologies is an AI-powered traffic control system that makes intersections safer, more efficient, and cheaper for cities to maintain. We use computer vision and commodity traffic cameras to build a 3D map of everything happening at the intersection in real time. We then use this map to simulate all possible signal states and pick the one that leads to the lowest wait times. Without solution, cities can

1. Greatly reduce traffic congestion
2. Eliminate the frustration of waiting at empty intersections
3. Respond to pedestrians and cyclists
4. Give priority to buses and emergency vehicles
5. And reduce resident complaints about their intersections

All in a single system, and all without ever requiring specialized hardware or bespoke traffic studies!

Ask

Know anyone working in municipal public works, transportation departments, or traffic engineering consulting firms? We’d love to meet them! founders@roundabout.tech

Team

Collin Barnwell is a software leader and engineer with extensive experience building and shipping safety-critical IoT and camera devices, including a device that has been implanted in humans. He's worked at Google and Verily Life Sciences.

Sabeek Pradhan is a computer vision expert with extensive experience building systems for high-accuracy vehicle, pedestrian and bike detection and tracking at the self-driving car companies Zoox, Waymo and Applied Intuition, and his software has been deployed on public roads in fully driverless vehicles.