Carnegie Mellon Tries Crowdsourcing To Develop Optimal Electric Car Formula

• By Dian Schaffhauser
• 08/09/10

Carnegie Mellon University has turned to crowdsourcing to develop new ideas for managing power in electric cars. Researchers at the Pittsburgh institution have announced a contest to find the most efficient methods, and the grand prize will be an electric car. The contest is a project of ChargeCar, a university-sponsored community effort for making electrical vehicle travel practical and affordable. That project, in turn, comes out of the Community Robotics, Education and Technology Empowerment (CREATE) Lab, a community effort to further human-robot interaction.

"The number of variables that could possibly affect an electric car's performance and the strain on its batteries is virtually infinite," said Illah Nourbakhsh, associate professor of robotics and director of the Robotics Institute's CREATE Lab. "Crowdsourcing is our best hope for sifting through those variables to find the optimal method for managing the flow of current between the motor and the power storage system. A contest seems the best way to draw a crowd and tap its wisdom." Crowdsourcing is a mechanism by which people can contribute their ideas or actions to a common goal.

People who want to participate in this particular effort download a software package from the ChargeCar Web site, as well as data files on driving behavior, and some examples of power management policies from the project Web site.

Alex Styler, a robotics graduate student supervising the contest, said that some basic knowledge of Java programming is required to develop an entry, which consists of a power management algorithm. "But even if you don't know anything about programming, you could learn what you need in a day," he said. "The most important quality for winning this contest is intuition, not programming skill."

The ultimate outcome of the contest is to find the most efficient way to handle the flow of power among the components of the electric car--the batteries, supercapacitors (for handling surges of power needed in driving activities), and motor. "Any source of information is fair game for people designing power management policies, as long as the source is publicly available," Styler said. All entries themselves will also be open source and available online for scrutiny by competitors.

The university said the contest will include driving histories from eight "secret drivers," uploaded monthly. Contestants who upload their algorithms to the contest will have their ideas tested against the data generated by those drivers. The top scorers will be listed on an online leader board; the monthly leader wins a $250 gift card; and the top four scorers will be eligible for the grand prize.

The contest could run as long as 18 months; but if the right formula surfaces before then, the university will notify contestants so that they can submit a final algorithm for judging. The ultimate prize--that electric car--may actually be built by the university itself if the competition runs long enough.