Michael (Mike) Montemerlo, TJ ’93, is a Senior Staff Software Engineer at Google working on self-driving cars. He received his B.S. and M.S. in Electrical/Computer Engineering (1997) and PhD in Robotics (2003) from Carnegie Mellon University, where he was a Hertz Fellow. After graduation, Sebastian Thrun, one of Montemerlo’s advisors for his PhD thesis on efficient probabilistic methods for mobile robots, recruited him to the Stanford Artificial Intelligence Lab. It was at Stanford that Montemerlo led the software team for Stanley, the autonomous Volkswagen that won the 2005 DARPA Grand Challenge.
Now Montemerlo, along with other DARPA Grand Challenge alumni, works on Google’s Self-Driving Car project, which Thrun started at Google in 2009. Initially a secret, the project has been public for several years, and in Silicon Valley it’s not unusual to see one of Google’s many test cars out in commuter traffic.
Driverless cars have the potential to be safer than cars driven by human drivers, and preliminary test results are promising. That improvement translates to lower mortality and morbidity rates, as well as increased independence for disabled or elderly adults. (Due to legal restrictions as well as the cars’ limitations, all driverless test vehicles currently on the road have human drivers behind the wheel ready to take over if needed). Once more such cars are on the road, highway efficiency should also improve as cars that sense each other can more safely and easily pack closer together, which in turn would reduce commuting time and emissions.
In January, Montemerlo was the keynote speaker for the Rae Dorough Speakers Series held at the Bankhead Theater in Livermore, California, where an ad for his talk read:
“Self-driving vehicles hold the promise of saving lives and reshaping our relationship with the automobile. The Google Self-Driving Car project was created to rapidly advance this technology. Mike will entertain you with the adventures that Google has had as he demonstrates the capabilities (and limitations) of these vehicles.”
Although several automotive manufacturers are simultaneously developing their own autonomous vehicle prototypes, Google’s Self-Driving Car depends on two main innovations. The first is LIDAR, an acronymn for light detection and ranging, which works on the same principle as radar and is used to create a 3-D model of the car’s surroundings, accurate to the centimeter. The second is Google Chauffeur, the software that interprets LIDAR’s data, which makes decisions about speed, direction, and distance from nearby cars and objects, performs constant diagnostic tests, and actually drives the car. Montemerlo works on algorithms for building the large maps the cars use to drive.
When Montemerlo attended TJ, the idea of driverless cars was a mere fantasy. “I fondly remember learning from Mr. Hyatt in the Computer Systems Lab many years ago,” he said, aware that mentioning a long-retired teacher would make him sound middle-aged. But today’s students, who are being mentored by some of the ablest high school faculty anywhere and who naturally dream of working on the next big thing, are thrilled to learn that someone who once stood in their shoes is making the future happen.