U.C. San Diego Unveils Improved Pedestrian Spotter for Cars

(TNS) -- UC San Diego has created experimental software that's meant to make it faster and easier for self-driving cars to detect pedestrians, a hot area of technology dominated by such major companies as Ford and Google.

The new system involves traditional cameras, which see pedestrians, and novel software that greatly refines the information, a form of machine learning.

"We're using cameras instead of sensors to identify pedestrians," said Nuno Vasconcelos, the UC San Diego electrical engineering professor who is leading the development team. "It's a lot cheaper way to create the detection system. In the end, this should be the better way to go. We're already working with a company that can test the system."

Vasconcelos' software program is designed to help operate pedestrian detection systems that run in near real-time (2-4 picture frames per second). It's part of a larger effort to enable robotic systems to spot and evaluate objects in their environment.

©2016 The San Diego Union-Tribune Distributed by Tribune Content Agency, LLC.