FALL, 2014 -- In a garage at the Richmond Field Station, about seven miles from the UC Berkeley campus, a late model Hyundai Azera with Michigan manufacturer plates is undergoing a radical transformation.
Tammo Zobel, a UC Berkeley researcher, is installing laser sensors, or Lidar, radar, cameras and an extraordinarily accurate GPS in its silver-gray body. He’s connecting these devices with more than a kilometer of additional wiring neatly packaged inside the vehicle.
Zobel, who has worked in the area of control theory and knows the mechanics of cars inside and out, explains that the hardware and software with which he’s equipping the Hyundai sedan will provide vital information needed to develop control systems that will revolutionize the safety and driver experience of future vehicles.
Benefits of partnership
Zobel and the grey Azera are just part of a much larger transportation research project at Berkeley, one that showcases the importance of university research for industry while offering students the opportunity to collaborate on cutting-edge science in control theory and control systems technology.
The Hyundai-Berkeley partnership is unique and unprecedented. Hyundai vehicles at UC Berkeley are used as a platform to demonstrate the effectiveness and the value of state of the art control theory and control technology.
“It is a great asset for our researchers and our collaboration,” said Borrelli.
In addition, several Hyundai engineers visit the university each year for training and to work closely with the Center of Excellence scholars on research projects under the guidance of experts like Hedrick and Borrelli.
“These visiting Hyundai engineers already have either their master’s or doctoral degrees, but here they have the opportunity to learn about new developments in the area of control and optimization of dynamic systems,” explained Hedrick. “Also, here they can sit in on classes as well as work in the Vehicle Dynamics and Control Lab
Borrelli and Hedrick visit Seoul often each year to meet the company’s research engineers, review research projects, and choose the most interesting research directions that would benefit both UC Berkeley graduate students and Hyundai. Examples of such projects include advanced vehicle dynamics controllers, self-parking vehicles, and control of four-wheel drive electric vehicles.
“Part of what we’re doing is transferring the theory and tools developed at Berkeley to Hyundai engineers, and part of what we’re doing with the funding Hyundai provides is to support our researchers and an associate director,” explains Hedrick. “It’s a wonderful relationship whose model we hope to extend to other automotive companies, just as many UC Berkeley centers are funded by several industrial partners on precompetitive research.”
Mimicking your driving personality
In the last decade, manufacturers have added a number of safety features, such as active steering, assisted parking systems and autonomous braking to avoid hitting pedestrians. Such advances are leading automobile manufacturers inexorably toward vehicles that no longer need a human driver.
As safety features are added they become more refined. One area of research that has attracted funding from government and several OEMs is work Borrelli has done: modeling the driver’s “style” of driving, so that when the vehicle takes over, it drives in a way that is comfortable and familiar to its human driver.
“The sensors have captured data about your pedal position, your acceleration, your steering, how you drive when you overtake another car, the distance you maintain behind another vehicle,” explained Borrelli, “so when the car takes over it’s a comfortable transition.”
“So a big difference between the Azera and existing fully-autonomous cars is this greater human-vehicle interaction. This car will drive like you do, not like Company X is driving the car,” he added.
“We are also using model-based methods for designing a self-driving vehicle as opposed to the usual industrial approach which involves a lot of manual coding. And a model-based approach has many advantages. The biggest may be maintenance: another engineer can easily interpret the control algorithm, and modify and implement it on a vehicle with different characteristics.”
How soon before safety features deliver an autonomous vehicle?
For some people improved safety features are great, but what they really want is a self-driving car.
Older people, who may be losing their ability to drive, hope self-driving cars will prolong the independence they treasure. Safety enthusiasts believe handing over control to our vehicles will reduce crashes and injuries. Transportation engineers think self-driving cars could expand the carrying capacity of our congested roadways. And some of us would rather read a book while our vehicles whisk us from one city to the next.
Despite the hype, it will be awhile.
Fleet vehicles may be the first to try it out, partly because they will be in a better position to deal with the question of liability in the event of an accident.
In terms of passenger cars, Hedrick believes humans will continue to be in charge for the next 10 years, with improving safety assistance from the vehicle.
By twenty years? Maybe then we can relax and enjoy a good novel or a nice long nap on the road between San Francisco and Los Angeles.