A $2,400 Class to Make Anyone a Self-Driving Car Engineer
Sure, the autonomous era will wipe out a lot of jobs. Automakers, tech titans, and startups are racing to essentially put four million truckers, cabbies and other drivers out of work. But like all radical technological shifts, self-driving cars will provide opportunities, too—for those with the right skills.
A college degree isn’t enough. Working in the most compelling part of this field requires an understanding of deep learning, the branch of artificial intelligence that trains computers to do things like discern pedestrians from lamp posts. Universities can’t crank out graduates fast enough.
To help meet demand, online educator Udacity is joining Mercedes-Benz, Nvidia, Chinese ridesharing behemoth Didi, and Otto, the autonomous truck outfit that Uber recently bought. They’re launching a 27-week course that promises to turn anyone with $2,400 and basic coding experience into a deep learning engineer.
Most of the major automakers and tech companies like Google and Baidu are racing to develop autonomous vehicles. Uber has robo-cars providing rides in Pittsburgh, Pennsylvania. Udacity’s “self-driving car engineer nanodegree” underscores the speed with which this change is coming.
A New Kind of Engineer
It’s a natural move for Udacity president Sebastian Thrun, who’s been studying artificial intelligence since the 1980s and competed in Darpa’s self-driving vehicle challenges in the early 2000s. He launched Google’s autonomous vehicle program before founding Udacity in 2014. He’ll start with 250 students, but “if you were to get 50,000 students on day one, I wouldn’t see any difficulties placing them in jobs”
Boston Global Consulting predicts that the autonomous driving market will hit $42 billion and create 100,000 new jobs by 2025. “The demand for great deep learning engineers is incredibly high,” says Richard Socher, who teaches the subject at Stanford and is Salesforce’s chief scientist. The demand far exceeds the supply. “There’s clearly a war for talent,” says Axel Gern, who runs Mercedes-Benz’s autonomous driving program in North America.
Udacity claims that more than 30,000 people have expressed interest in the program. You can see why. Udacity says base salaries for self-driving car engineers range from $66,800 to $210,000, and its four partners have agreed to hire the sharpest graduates.
Candidates must have solid programming skills in Python, C++, or another coding language, and knowledge of statistics, algebra, and calculus. Students move at their own pace through the three nine-week trimesters, but should expect to spend 10 hours a week on coursework. They will work through tasks required to make a car drive itself: using camera images to detect lane lines, training a car to determine its location without GPS, path planning, and so on. At the end of the course, students will have the opportunity to work on a car on a test track.
Mercedes, Nvidia, and Otto will provide students with real-world problems, and insights from its own engineers. Students will receive prompt evaluations—not grades—of their work by experts in the field. The idea to to train students, not rate them.
Jianxiong Xiao, who ran Princeton University’s computer vision and robotics lab until leaving to start a robo-car startup, says anyone with a solid foundation in programming will come out with a strong understanding of deep learning. “If you want a basic understanding, it’s not that hard,” he says. And “for a big company, if they want to hire so many people, this is probably the only choice.”
Thrun hopes to diversify the pool of candidates, too. Udacity’s scholarship programs in Syria, India, Egypt, and elsewhere will bring in otherwise unnoticed talent. That doesn’t just benefit Udacity’s students—it could make for better products.
“Creative solutions come from doing creative things,” says Anthony Levandowski, a co-founder of Otto who worked with Thrun on Google’s self-driving car. Bringing in people from different backgrounds makes you think differently about how you solve problems. And so the advent of autonomy moves another step closer.
Read this article: