Once a physics teacher in the Navy, now a ‘car guy’ at Stanford

John Alsterda, a doctoral student who is developing a new emergency avoidance controller for self-driving cars, is one of 91 military service veterans studying at Stanford.

As John Alsterda neared the end of his assignment at the U.S. Navy’s Nuclear Power School in South Carolina, he decided it was time to pursue his dream of earning a doctorate.

John Alsterda with the X1, an autonomous experimental car that was designed and built by Stanford PhD students. (Image credit: Farrin Abbott)

Initially, he planned to study nuclear engineering.

But Alsterda changed course after watching a self-driving coupe named Shelley zip around a racetrack at 120 mph in a video published by the Dynamic Design Lab, which is part of the Mechanical Engineering Department in Stanford’s School of Engineering.

The lab, led by Chris Gerdes, a professor of mechanical engineering and director of the Center for Automotive Research at Stanford (CARS), studies how cars move, how humans drive cars and how to design future cars that will work cooperatively with the driver or drive themselves. Alsterda was thrilled when he was admitted to the graduate program in mechanical engineering in 2015. He said it has been a transformative experience.

“I was never a ‘car guy’ before starting the program,” said Alsterda, whose first step on the path to a doctorate was earning a master’s degree in mechanical engineering at Stanford – which he achieved in 2017. “Now I’m about as thick into cars as a person could be.”

As a PhD student, Alsterda is developing an algorithm designed to teach self-driving cars to anticipate events that might take place, instead of reacting when emergencies occur – just like good human drivers.

From the Navy to the Farm

Alsterda, who grew up in Chicago, earned a bachelor’s degree in physics at the University of Illinois at Urbana-Champaign. Inspired by his high school physics teacher and his early experience as an undergraduate, Alsterda decided to teach physics.

The Navy recruited Alsterda in college, promising him a commission as a lieutenant and a teaching position after he graduated.

He spent four years at the Naval Nuclear Power School, where he taught physics to enlisted men and women, and reactor dynamics and core characteristics to officers. Eventually, he became director of the school’s math and physics division.

Alsterda, who serves in the reserve of the Office of Naval Research, is one of 91 veterans studying at Stanford, including undergraduate and graduate students, and visiting fellows at the Center for International Security and Cooperation and the Hoover Institution.

Working on our driverless future

Every six weeks or so, Alsterda accompanies an autonomous experimental car known as the X1 to a racetrack in northern California to test his emergency avoidance algorithms. The car, designed and built by Stanford PhD students more than a decade ago, has been maintained and upgraded by subsequent generations of students, including Alsterda.

His research goal is to develop the next generation of vehicle control systems for driverless cars. It is a field that has attracted many researchers.

“What makes my work unique is the contingency planning aspect,” Alsterda said. “A contingency is a potential emergency that we’ve identified. The emergency might be an ice slick on a winter day or a child walking on the sidewalk who might dash into the street. The important thing here is that the event hasn’t happened yet, and we don’t know if it will.”

Alsterda said it would be easy to slow the car down to a crawl to avoid such emergencies, but a vehicle that slows down for every uncertainty would be a nuisance on the road, and possibly even dangerous.

“So my algorithm asks, How well can we maximize our performance objectives – traveling down streets smoothly and quickly – while guaranteeing the potential emergency can be safely avoided,” he said. “My algorithm does this by maintaining a contingency plan – an alternate trajectory to avert an identified potential emergency. As long as this contingency plan exists, my algorithm can concentrate on driving expeditiously to its destination.”

In July 2019, Alsterda co-authored a paper describing how his fledgling controller safely steered an autonomous car through an icy left turn at a research facility in Sweden. His current goal is to incorporate throttle and brakes as well, to safely navigate on ice and through other contingencies using a car’s full capabilities. He will return to Sweden in February to test the latest iteration of his controller.

“I’m so lucky to be here at Stanford with the opportunity to do cutting-edge research,” Alsterda said. “The expert community and resources I need are right here at my fingertips. I hope I can rise to the challenge and build something to fundamentally improve vehicle and robotic safety.”