Ford gathers street-smarts to guide cars of future
Mcity urban test facility is bustling with autonomous vehicles including a pair of driverless shuttles that will transport students on the North Campus as well as the latest from Ford. Todd McInturf, The Detroit News
Ann Arbor — The University of Michigan’s one-of-a-kind research center for autonomous and connected vehicles is something like a practice field for Ford Motor Co.
Inside that controlled environment, Ford runs drills that put the automaker’s fully autonomous Fusion Hybrid test vehicles in considerably dangerous — yet scripted — situations, like unexpected pedestrians crossing the road, or bicyclists riding in the street. The objective is to ensure the vehicle’s sensors, cameras and mapping system all react properly to anything it could encounter on the road.
Collected data will be used in future autonomous cars, Ford said.
That research code will be incorporated into the “brain” of a fully autonomous vehicle without a steering wheel, accelerator or brake pedal that Ford plans to bring to market by 2021. Ford handed the code it had developed over 10 years of autonomous research to artificial intelligence company Argo AI earlier this year, shortly after the Dearborn-based automaker announced it would invest $1 billion in the company.
“We are showing and testing out our research code,” Ken Washington, Ford vice president of research and advanced engineering and chief technology officer, said Wednesday. “We’re taking the code that we gave (Argo AI) to start from, and we’re using that as the base for how to explore new capabilities, new things that we can bring to the future. (Mcity) is the perfect place to do that. It simulates a city environment. You need to have impromptu, real-world experiences, but you also need to have scripted environments so you can say ‘I need to test against this scenario.’ ”
The initial code is just a starting point for autonomous vehicles, Washington and members of his team gathered here said Wednesday. The hardware will change, and there is always new data to be harvested and fed back into the autonomous systems. And there are many, many drills to be done.
Last fall, Ford demonstrated a fully autonomous Fusion Hybrid on public roads in Dearborn around Ford’s Product Development Center. The automaker also tests the vehicles on public roads in California, Arizona and its gated campus in Dearborn.
The Mcity work supplements that.
“This becomes the foundation for what we could offer to Argo both ongoing and later (for) the next generation,” Washington said.
The Wednesday afternoon test drive around part of the 32-acre facility demonstrated Ford’s “baseline” technology, according to Washington.
Codrin Cionca, a Dearborn-based Ford research scientist, sat behind the steering wheel of one of the white Fusions to take over in case of an emergency. His hands were off the wheel and his feet off the pedals. Beside him, Ford Research Engineer Helen Kourous monitored the vehicle’s location on a map of the test track, also uploaded into the vehicle’s computer system.
The autonomous car used data collected live from a system of four lidar sensors, seven cameras and a series of radar sensors around the car. It essentially laid that information over the map, Kourous said. As the vehicle followed lanes and obeyed stop signs and yield signs already programmed into its brain, its cameras were feeding information on traffic light signals — red, yellow and green — other vehicles on the road, activity at intersections and pedestrians crossing the road.
And when a bicyclist unexpected pulled into the street in front of the vehicle, the Fusion detected an object. Its cameras and sensors determined that object to be a human on a bike, and the car crept along slowly multiple car-lengths behind the man until he got out of the street. Then it proceeded along through the intersection at 15 to 20 mph.
For now, Ford is relying on some variation of that system of sensors and cameras to be the “eyes” of the autonomous vehicle. Jim McBride, senior technical leader for autonomous vehicles, and Randy Visintainer, director for autonomous vehicles, said although the technology is emerging that will allow vehicles to “talk” to each other or stop lights and other infrastructure, the first phase of autonomous vehicles need to be able to operate safely and efficiently without that technology, because it won’t be the standard for a long time.
Ford plans to add those sensors to supplement the array it currently uses on the autonomous vehicles as they become more common.
Washington, who recently was appointed to chief technology officer under Ford’s new management structure, said the biggest challenge as Ford brings its first round of autonomous vehicles to production is to use its “powerful” position as an automaker and a technology company to manufacture a strong, safe product.
“We feel good about our position,” he said. “The biggest challenge for us (going forward) is helping the public and the policy makers understand what’s real and what’s fake” about how autonomous vehicles operate.