Cirrus Traffic - The idea

Cities should use groups of UAVs to study traffic, provide useful real-time information about traffic, and find and ticket dangerous drivers. There are already UAVs that stay airborne for two weeks at a time. These low maintenance UAVs are perfect for this task. The UAVs would have a payload of camera equipment and communication equipment to broadcast (encrypted, of course) its feed. It would use solar panels to run during the day, and at night, it would turn off the camera and run on battery power so that it wouldn’t have to use energy on landing and takeoff every day.

The camera spec

The primary use case is tracking highway traffic. We can assume cars may be moving at up to 80 mph (36 m/s). Using a Honda Accord as an example car, that means that the car travels 7.5 car lengths per second (the Accord is 4.8 meters long). For the images of cars to not be blurry then the frame rate must be much faster than 7.5 fps (or at least the exposure time must be much less than 130 milliseconds). It’s not uncommon for modern digital cameras to be able to capture 60 fps at 1080p. Unfortunately, that won’t be high enough resolution, as each pixel would cover several cars. Sacrificing frame rate, it’s possible to get 6fps and an exposure time much less than 130 milliseconds with the Sony Alpha SLT-A99.

The UAVs would fly at an altitude between 3,000 and 10,000 meters, which is high enough to capture large chucks of the city. The amount of area covered by each UAV won’t be limited by the altitude, but instead by the resolution of images recorded. Each car on the road should be covered by several pixels. A reasonable scale is each pixel is a meter by a meter. That would make cars around 5 pixels long. The sensor of the SLT-A99 is 6000 x 3376 pixels, so it could cover 20 square kilometers. With that amount of coverage, it would only take 65 UAVs to cover all of Los Angeles, which is not a particularly small city. With a little engineering effort, one could combine several of the image sensors onto a single UAV and cut the number of UAVs needed to patrol the skies.

What to do with the data

With detailed data about the routes of every car and their reaction to their surroundings, it should be possible to make very accurate models of driver behavior. These models could easily be validated against new days of traffic. These models could help answer what changes to road design and construction could reduce traffic accidents and jams. It could also give insights into what makes for good defensive driving and traffic reducing driving. What if you could alert drivers that there would be a jam ahead and convince them that they should slow down by 10 mph? Would that help or hinder? There are many interesting questions that could be answered with these data.

Getting dangerous drivers off the road

The data from the UAVs could find the most aggressive drivers and send police cars to pull them over or combine the UAV data with highway camera images and just mail the offender a ticket.