Blocking the bus lane? Cities are using AI to find and fine you

Matthew Carreiro // Shutterstock Think you can pull over and park in a bus lane to grab your morning latte? Think again–at least in cities using computer vision technology to catch and fine drivers who block bus lanes. As cities across the U.S. invest in new bus lane infrastructure to speed up travel times and prevent delays, drivers continue to flout the red (or green) paint, Next City reports. One 2023 study from Philadelphia’s SEPTA recorded 36,000 violations during a 70-day pilot on two bus lines, with half of all bus stops obstructed at least once and 10,000 riders delayed. That’s a problem. Enter Hayden AI, a San Francisco-based startup deploying bus-mounted cameras that are programmed to capture and record vehicles blocking bus lanes. So far, cities using Hayden’s technology include D.C., New York, Oakland, and Los Angeles. Other cities are in the process of procuring similar technology. In August, the Philadelphia Parking Authority put out an RFP for a bus camera enforcement system. “If a city is going to spend millions of dollars redesigning its network to add these lanes, there was some realization or recognition that they had to enforce the use of those lanes,” says Charley Territo, chief growth officer for Hayden. Using AI makes enforcement less time and labor-intensive, according to the startup. And there’s a certain poetic justice to buses themselves as tools of enforcement. “If you get a ticket, you were caught actually blocking a bus, since it’s the bus that took the picture,” says Jenna Fortunati, the company’s spokesperson. How the Technology Works Before starting a pilot, Hayden AI maps out details of the bus route, including the location and dimensions of bus stops, the lane lines, and more. It then programs algorithms so that onboard cameras can detect when a vehicle is obstructing the bus lane. The system uses two cameras mounted behind the windshield. One camera, called the context camera, “watches” for violations. Once the context camera detects an offending vehicle, it starts recording a 10-second evidence package. The second camera takes a picture of the vehicle’s license plate. The evidence package is saved to a computer on board, then sent to the cloud to be reviewed by the parking enforcement agency. Every evidence package is reviewed by a human, not a computer. As the system learns the route, machine learning makes it even more accurate at detecting bus lane obstructions. Why Ticket Drivers Who Block Bus Lanes? Automated enforcement, such as red light cameras and bus lane enforcement, isn’t always popular with drivers–for obvious reasons. However, advocates say ticketing violators is good for bus riders, especially people with disabilities. If a car parks at a bus stop, it blocks the bus from pulling over to the curb. To board people using wheelchairs, the bus must lower the ramp to the street at a dangerous angle. In these cases, people who use mobility devices can’t safely board–or board at all. “Some wheelchairs may have backpacks that have essential life needs for a person in the wheelchair, like a feeder, or they may have a colostomy bag on the back of the chair, which could cause the chair to be heavy at the top,” explains Kwanesha Clarke, an independent living specialist at Liberty Resources, a Philadelphia-based independent living center partnering with Hayden AI. “When bus lanes are blocked, it stops people in wheelchairs from participating in their community,” she adds. “It stops us from going to school. It stops us from achieving those personal and educational goals and even our recreational enjoyment with friends and family.” So far, bus lane enforcement pilots show success in speeding up buses and increasing safety for road users. New York’s MTA reports that its enforcement program has increased bus speeds by 5%, reduced collisions by 20% and reduced emissions by up to 10%. In addition, over 90% of drivers who receive a ticket don’t receive a second ticket. What About Privacy? With cameras on the bus capturing video of what’s going on outside the bus, there are concerns about privacy. Unlike other systems, Hayden AI’s model doesn’t continuously record data or stream that data to the cloud. Although the context camera is always “watching,” it does not record or save video except when a vehicle is detected blocking a bus lane. The company says it has the ability to blur faces and obscure other license plates, technology being requested by European cities. Hayden started deploying in Europe this year, with pilots in Gda sk, Poland; Braga, Portugal; Tallinn, Estonia; and Barcelona, Spain. However, in the U.S., transit agencies have pushed back on these privacy features–though there might be a shift. “We were always prevented, saying that this would amount to the tampering of evidence,” says Vaibhav Ghadiok, co-founder and chief technology officer at Hayden. “As we’ve expanded into California, there is, let’s call it, a greater interest in such technology … But I think Europe leads the way at this point.” Could Bike Lane Enforcement Be Next? Computer vision technology could capture more than just vehicles blocking bus lanes. In June, the New York MTA announced that it was expanding its bus lane enforcement pilot to include double parking and bus stop violations. “There is a demand to enforce paratransit bus stops, bike lanes, vehicles moving in bus lanes, enforcing loading zones,” says Ghadiok. “There’s this whole sort of cornucopia of use cases there to make a city more efficient.”Hayden also sees potential in the future to inventory and monitor street-level infrastructure like bus shelters and curbs. What if a bus could “see” a bus shelter that needs maintenance, for example, and send that information to the appropriate department? “There may be broken glass, there may be graffiti on the bus stop,” says Ghadiok. “Those are things where a city can be immediately alerted and fixed.” This story was produced through our Equitable Cities Fellowship for Social Impact Design, which is made possible with funding from the National Endowment for the Arts. This story was produced by Next City, a nonprofit newsroom covering solutions for equitable cities, and reviewed and distributed by Stacker.