Engineers look at bridges differently than the rest of us. When Jim Nickolaou, an engineer at GM working on advanced safety systems, looks at the Brooklyn Bridge, he admires the structure for its beauty and as a feat of engineering. But he also considers the iconic structure a perfect test bed for the latest driver assist systems on the Cadillac XTS and upcoming ATS.
Nickolaou oversees a team that tests what’s known as “sensor fusion” to see how radar and cameras systems – part of a car’s driver aids – work together. “Sensor fusion has been around awhile on the military side, where you use the best from the different modalities of the various sensors to try and produce results that’s better than just one sensor,” Nickolaou, who worked on radar systems for F14 and F15 fighters jets before joining GM, told Wired. “Radar is good at finding moving objects and finding the range of moving objects through the Doppler shift. Cameras, on the other hand, are better at doing azimuth, or left to right, recognition of those same objects. So what we try to do is take the advantage of both types of sensors.”
Part of the idea behind sensor fusion testing is to intentionally try to confuse the radar and cameras systems that activate a car’s driver assist features such as Automatic Collision Preparation and Front and Rear Automatic Braking. And metal structures like bridges, with their girders and tightly packed traffic are perfect for vetting radar sensors, while cameras have to deal with varying light.
The Brooklyn Bridge isn’t the only one the team included in the tests. But it offers unique challenges to sensors, Nickolaou says, because of the metal super structure, guard rails, walkways and the opposing traffic flow.
“We drove the Golden Gate in San Francisco,” he recalls. “We crisscrossed the country from Portland and Seattle down to Louisiana, trying to locate what we believe are the most challenging locations for these sensors. But the Brooklyn Bridge epitomizes the challenges of radar energy bouncing off of the metal girders. There are also a lot of right angles on the girders and it tends to propagate the radio emissions from the radar back in with some information that might not be accurate. And the overhead girders throw shadows for the cameras. So there are a lot of challenges from that one bridge.”
And the team also went underground.
“As you dive into tunnels you get the light transmissions,” Nickolaou adds. “So we also covered all of the tunnels in New York.” They also uncovered a unique challenge for cameras in a tunnel in Pittsburgh, Pennsylvania. “There’s one tunnel that has tile on the sides,” Nickolaou notes. “The first time we went through it we found no issues with the sensors. The next time it was raining, and over time the water tracks went into the tunnel. And as people hit their brakes, brake lights were being reflected on the wall of the tunnel.” The camera may see it as a car braking ahead, Nickolaou explains. “That’s where the strength of the radar comes in; it knows it’s not a car because it sees the energy from the entire tunnel.”
Nickolaou admits that he doesn’t just view a bridge from an engineering or testing standpoint, and that like many others he also relates to a span’s aesthetic appeal. “Here in Michigan we have the Mackinac Bridge, and it’s really gorgeous, and the Golden Gate is really gorgeous,” he says. But the Brooklyn Bridge wins out on all counts. “We used the Brooklyn Bridge as one of our main locations we revisited over and over in the development process.” And, with its dramatic view of Manhattan, it’s his favorite from a visual standpoint too.
Source: Wired
Tidak ada komentar