The light that turned green shortly after 7 a.m. gave motorists on Flagler Street the legal right to proceed through the intersection, but not the ability. Stalled cars and trucks in front of them on Le Jeune Avenue made progress impossible. The congestion came from a disabled sedan blocking the right lane on Le Jeune about eight blocks away.
Already at work in a windowless office outside Doral, Carlos R. Vasquez peered into his video screen and tried to do something about the traffic mess.
“Northbound is backed up,” the county traffic engineer said, using the arrow buttons to control a traffic camera suspended over Le Jeune. “I’m going to go ahead and give it more green.”
With a few clicks, Vasquez extended the green light at Le Jeune and Seventh Street by 20 seconds, giving northbound motorists more time to clear the right-lane bottleneck.
Never miss a local story.
The keystrokes were the latest maneuver in Miami-Dade’s recent efforts to better choreograph green lights throughout the county’s nearly 3,000 intersections.
After years of having to get in a car to see traffic conditions around the lights they control, county engineers got their first remote-controlled camera over an intersection in late 2015. About 120 intersections have the cameras now, allowing the kind of workaround Vasquez implemented for the stalled sedan on Le Jeune from his terminal at the county’s Traffic Management Center off 36th Street near Doral.
“What we’re trying to do is adjust as quickly as possible,” said Frank Aira, chief of the county’s Traffic Signals and Signs Division. “In the past, we didn’t have this. We couldn’t even see or look. We now have the ability to make an adjustment. It may not be perfect. But it improves it a little bit.”
The computer is going to adjust and it’s going to adjust and it’s going to adjust. … It’s that continuous improvement 24 hours a day, at a scale we just couldn’t do with people.
Alice Bravo, Miami-Dade transportation director
Even with the new cameras, Miami-Dade plans to largely take real-time human decisions out of the picture. After testing computerized lights in 2016 and 2017, the county wants to begin expanding the technology to 10 of the busiest corridors by the end of the year.
The $12 million effort would turn over light-switching decisions to a computer system that uses a mix of sensors and algorithms to determine which part of the intersection should be given the green light at any given time. Digital analysis of a video feed replaces the human eye, while sensors pick up passing Bluetooth signals from vehicles to detect changes in average speeds.
From there, a computerized system coordinates changes in multiple intersections to try and speed traffic through the busiest corridors.
Miami-Dade started a test of the technology in November on 36th Street, with computers controlling eight intersections between 71st and 84th avenues. The computers allow the kind of rapid-fire calculations no human could manage. Shortly after Vasquez added 30 seconds to a green light to try and clear Le Jeune, the computerized lights on 36th Avenue sensed trouble — and took action, too. Their adjustments were just far subtler.
At 36th Street and 82nd Avenue, the computer took 16 seconds away from the eastbound left turn, and transferred those 16 seconds to the westbound lanes. Northbound motorists at 36th and Milam Dairy Road lost three seconds of green from the northbound lanes, while southbound motorists gained three seconds. Where 36th meets the Palmetto, five seconds went to the westbound lanes and five seconds were shaved off the exit ramp onto the expressway.
“The computer is going to adjust and it’s going to adjust and it’s going to adjust,” Alice Bravo, the county’s transportation director, said. “And it won’t do that just when there’s an incident. Today that travel speed might be 20 miles per hour. So it adjusts to help you reach the green light. The next day that speed might be 40 miles per hour. So it adjusts.”
“It’s that continuous improvement 24 hours a day,” she said, “at a scale we just couldn’t do with people.”
Miami-Dade is rushing to be part of the “adaptive light” movement, the term for computerized signals that automatically adjust to traffic conditions. The county’s transportation department wants to give a no-bid contract to the California company that ran the test program, Econolite, in order to speed installations. Econolite’s pending contract won approval from the County Commission’s transportation committee on Thursday, with a final vote expected in July.
The pace to award the $11.8 million, two-year contract caught the eye of the county’s Office of the Inspector General, which issued a June 12 report noting some downsides to the proposal.
One problem is Miami-Dade requires that before a company bids on such a contract, county personnel must test and approve its devices — even if its equipment already has been certified at the state or national level. Because the county has only one staffer assigned to testing smart traffic signals, Econolite currently has Miami-Dade’s planned $160 million light upgrade to itself.
“Despite the best efforts of staff,” according to the report, “only one manufacturer of the [adaptive] traffic controller has been locally approved and certified.”
An Econolite competitor, McCain, also out of California, is undergoing county testing and should be ready to compete for the larger adaptive-signal contract that Miami-Dade plans after Econolite’s installation on the 10 busiest corridors.
While elected leaders are touting a multi-billion-dollar plan to upgrade public transit countywide, the biggest gains on commuting times come on the road. In Miami-Dade, of the people who travel to work, nearly 90 percent drive, according to county statistics.
That’s made traffic-light synchronization a buzzword for years as drivers press for relief from clogged intersections. The synchronization process for most lights still relies on pre-set schedules based on traffic assumptions for various times of the day. While county engineers can reprogram the lights from headquarters, they were unable to monitor traffic conditions before the intersection cameras arrived.
Even so, adjusting the schedules can yield results. After a consultant recommended changes to the automatic light schedules on U.S. 1, Miami-Dade says northbound travel times decreased an average of 10 minutes between Southwest 152nd Street and Southwest 16th Avenue.
The Econolite contract includes nearly 700 video sensors, which allow the computer to analyze not only traffic volume but also detect if cars waiting for a left-hand green light are backing up into the main lanes. The county’s plan is to let computers take over the real-time decisions with lights, rather than beef up its command-center staff.
“We have 11 engineers right now trying to see all 3,000 intersections,” said Aira, the signals chief. “From a human standpoint, you’d never have enough engineers to monitor everything.”
Commuters may have trouble detecting the computer’s work. An Econolite analysis of its test on 36th Street at the start of the year found it was able to speed travel time through the mile-long corridor by as much as 19 percent. But that amounted to only a 49-second improvement for motorists.
“The regular commuter is not going to say: ‘Oh wow, this is completely different,’ ” said Aleksandar Stevanovic, an engineering professor at Florida Atlantic University and director of the school’s Lab for Adaptive Traffic Operations and Management.
But Stevanovic said tweaking an intersection to allow an extra 10 vehicles through a green light on a main road can bring subtle improvements for thousands of drivers if it leads to fewer jammed intersections at rush hour. “You’ll see real improvement in capacity — being able to service more vehicles and making sure gridlock does not spread through the network,” he said.
For Vasquez, that was how he measured success in trying to give vehicles more time to get around the stalled sedan on Le Jeune.
“You see how backed up it is? It’s really bad,” he said, pointing to the screen showing vehicles crawling forward as they awaited an opening in the two lanes of traffic still moving. A tow truck was there, along with a traffic-patrol car. “It’s not getting better because they’re still on the side.”
He said there is really no pattern of light changing — no extended green light, or quick red — that could relieve the gridlock of a blocked lane during morning rush hour. The best he can do is try to speed the crawl enough to prevent other intersections on Le Jeune from being blocked, too. But the extra green can’t last too long, or the cross street gets too clogged.
“We can’t be too aggressive,” he said. “This is a very congested intersection.”