When is a car not a car? When it’s a network.
That’s how the University of Cincinnati’s Jiaqi Ma describes the coming driverless revolution.
Driverless cars will be self-sufficient enough to respond independently to changing traffic conditions or obstacles. But Ma also sees them as part of a bigger network, connected both to other nearby vehicles and the intersections of roads around them.
Ma, an assistant professor of civil engineering, is studying driverless technology at UC’s College of Engineering and Applied Science
. The goal is to get an autonomous car’s sensors and artificial intelligence to exceed the National Highway Traffic Safety Administration’s current driving safety standards in every way, he said.
“That’s a requirement. If a human driver can do it, we should be able to do it with a driverless car,” he said. “We’re far less tolerant of errors in autonomous systems. So we want to reach the level where human beings can trust the system and reach people’s expectations.”
Google’s autonomous car development company Waymo this year is launching a self-driving fleet of taxis in the United States. Google says its digital chauffeurs have driven 4 million miles and counting. Its minivan cabs are operating in test cities in five states.
Tesla said it is installing partially autonomous hardware in all of its new vehicles, even if it might be some time before all functions can be deployed. This includes a suite of eight cameras that can see up to 800 feet ahead, 12 ultrasonic sensors that detect hard and soft objects and forward-facing radar that can see through rain, fog or dust.
The California automaker is advertising a “smart summon” feature in which the vehicle serves as its own valet.
General Motors’ Super Cruise offers similar features for use on highways, advertised as “the world’s first hands-free driving system for the freeway.” Super Cruise illuminates a lightbar on the steering wheel when the hands-free function is engaged. But this is hardly autopilot. The driver is still expected to pay attention to the road.
Car engineers use a scale from 0 to 5 to describe the level of autonomy provided by vehicles. Zero means the car is controlled entirely by the driver. Level 5 is fully autonomous. General Motors’ Super Cruise falls into Level 3 in which the driver must be ready to take the wheel at any time.
Autonomous technology is likely to be deployed in stages, which means driverless cars will share the road with human drivers. But UC’s Ma envisions a time not too far in the future when enough driverless vehicles will be on the road to work together as an efficient “platoon” that coordinates its movements.
Anyone who has waited to turn at a short traffic light is familiar with the concept. When the arrow turns green, the first driver takes a second or longer to pull forward. The second driver waits a second or more after that and so on, allowing just a few vehicles to make the light.
Ma is working on a research project with the U.S. Department of Transportation called Eco Drive in which driverless cars will be able to communicate with each other and traffic lights. When the light turns green, an entire platoon of vehicles will move virtually in tandem, allowing many more vehicles to proceed.
Likewise, these connected and automated vehicles will know when an obstacle ahead prompts a slowdown, which could prevent emergency braking or the traffic phenomenon known as shock waves. These phantom obstacles occur whenever someone in heavy highway traffic brakes, creating a ripple effect that can last long after the original obstacle has vanished.
“These cascading effects are virtually unavoidable with human drivers,” Ma said. “In one study, they asked 20 vehicles to drive in a circle. And eventually, they all ended up in stop and go traffic. Everyone has different perceptual reactions.”
Autonomous vehicles will share data with each other and sensors in traffic controls every tenth of a second, vastly improving reaction time. Ma developed an algorithm for vehicles in platoons that improved fuel efficiency by as much as 20 percent by smoothing the stop-and-start traffic.
“We proved our algorithm was effective,” he said.
Cities such as Cincinnati are embracing these technologies to improve public transportation, said Zack Huhn, president of the consulting firm Venture Smarter. He envisions autonomous public shuttles being introduced first to places such as the Cincinnati/Northern Kentucky International Airport or even UC’s campuses.
“We’re really talking about integrating intelligent transportation solutions into our infrastructure and our public transit systems,” he said.
Huhn said UC will play an important role in helping to prove the feasibility, safety and long-term cost-effectiveness of these integrated systems.
“UC is deeply involved in making these things happen. It’s great for the region that this research lives here,” he said.
Meanwhile, engineers will have to overcome uncertainties such as the weather. Manufacturers are doing most of their testing in California and Arizona, sunny states, before testing the technology in the rain or snow. They want to perfect the system under ideal conditions first and then gradually introduce adverse conditions, Ma said.
“There are so many different scenarios of driving in the real world. They might be able to handle 90 percent or more of these driving scenarios, but that last 5 to 10 percent is a big challenge,” Ma said.
Ma also is helping the U.S. Department of Transportation conduct analysis, modeling and simulation studies of connected and automated vehicles. These computer models and simulations help researchers understand the effectiveness of different automation applications and networking tools before they undergo tests in the field, Ma said.
Simulations can help researchers examine how connected and automated vehicles and smart signals interact with virtual traffic, which could validate their research before field tests. It also allows researchers to isolate particular pieces of technology to adjust or refine the system.
He plans to work this year with the UC Office of Research
to purchase a test vehicle to study autonomous sensing and controls. Ma is collaborating with UC engineering professor Ou Ma (no relation).
Ou Ma said car sensors will have to deal with a host of challenges, from variations in light to high-contrast shadows.
“A solution is to add more, redundant sensors so if one fails, the others work,” he said. “You can always combine information from various sensors.”
UC's research holds promise to improve the quality of life for people in a number of ways, said Jennifer Krivickas, assistant vice president for integrated research at UC’s Office of Research
“Eventually, automated vehicles will provide people with safe, equitable and cost-effective access to education, jobs and quality health care,” Krivickas said.
“The advanced technologies and robotics such as those that Dr. Ou Ma is researching and developing will have both direct and indirect impacts on the economy, the environment and society.”
Government agencies will have to work with utilities to determine who pays for the technology upgrades to make traffic signals smart.
Initially, these systems will be more expensive, Ou Ma said. But the cost will come down over time as more manufacturers embrace these features.
“In the beginning, it’s always expensive. But the benefit is big. It’s worth the investment,” Ou Ma said.
Autonomous vehicles could have profound effects on the way we live and even where we live, he said.
“When we buy a house, we have to worry about the daily commute. It takes an hour to drive to work. But if you have automated vehicles, that’s another hour I could work in the car. Then I think cities could be spread over greater distances,” he said. “It’s going to be a big change in lifestyle.”
By Michael Miller
Photos by Joseph Fuqua II/UC Creative Services
Mar 19, 2018