Driving involves doing a lot of stuff, all at once, that you probably don’t think about.
Drivers — theoretically, anyway — have to maintain a certain follow distance, brake in a way that doesn’t cause road rage and stay in their lane. All of this, to be sure, takes a bit of practice.
Now imagine teaching a machine to do all that. In a car that you didn’t pick. And in a way that leaves the car street legal.
That is the challenge faced by members of Mississippi State University’s EcoCAR team, who not only have to struggle with those autonomy issues, but also with attempts to make vehicles more energy-efficient.
Vance Hudson, Jagdeo Singh and their teammates just returned from Arizona, where they competed to make a semi-autonomous Chevy Blazer run through the desert. Tuesday they told the Rotary Club of Columbus about the program and some of their engineering challenges.
Singh explained what, exactly, they perpetrated on the Blazer before heading to the desert.
“We replaced the stock engine with a GM two-liter engine with an automatic transmission to drive the front wheels,” he said. “In the back we added in a custom battery pack, as well as an electric motor and a team-designed gearbox to drive the rear wheels.”
The team also added a host of tech, including radar, cameras and a special antenna that could talk to traffic signals via Wi-Fi, he said.
“The antenna is basically a large Wi-Fi router,” Hudson said. “If you have smart traffic lights, they’ll broadcast that they’re turning red in 30 seconds, so you can plan what your vehicle’s going to do accordingly.”
The radar and camera look at the road in front of the vehicle, he said. The radar is really good at telling where something is, and calculating its velocity and direction. It’s not so great at lanes. Cameras, on the other hand, are good at exactly that. The problem, then, is getting the machine to combine those inputs and get a realistic picture of what’s ahead.
“Our team had to develop an algorithm to combine the data from the camera and the radar,” Hudson said, “and figure out that if the camera says something is 100 meters away, and the radar says 95, the radar is probably right. But the camera is probably right that it’s in our lane. (The algorithm) had to combine all that data to tell us what it’s doing.”
Once the information is in hand, there’s a whole other set of algorithms that had to be written to decide how to use that information, Hudson said.
“If you’re following a set speed and you’re approaching a car in front of you faster than it’s moving away, then you have to slow down and make sure we do it smoothly,” he said. “It will actually bring the (Blazer) all the way to a stop behind the other vehicle and then resume autonomously back up to the set speed.”
All of that data, in turn, was fed back into a fuel system to try to figure out how to improve fuel economy during all that stopping, starting and driving, he said.
“Most of the systems in your car just do it to make your life easier as a driver,” Hudson said. “We have to do it to improve fuel economy, as well.”
All of those systems — the various computers, sensors, batteries — also had to work in the Arizona heat.
Hudson said the Blazer had to make it through a 177-mile course in 110-degree heat. And it did, but not without a few hiccups.
“The air-conditioning system actually broke down,” he said. “We were driving with the windows down until the camera car got close, then we rolled them back up and acted like nothing was wrong.”
Hudson said all the competitors this year made it through the entire route successfully, which was an oddity. MSU’s team finished fifth overall.
When asked if he thought user-driven, individually owned cars would disappear in favor of autonomous short-term rentals, Hudson said probably not.
“I don’t think it’s likely in the U.S., just because of the culture of how our world was set up,” he said. “… Even with (electric vehicles) I think GM and the other Big Three are still planning on selling to private consumers directly.”
Brian Jones is the local government reporter for Columbus and Lowndes County.
