Home/Uncategorized/Argo AI collaborates with CMU, Georgia Tech researchers on self-driving cars

Argo AI collaborates with CMU, Georgia Tech researchers on self-driving cars


Aaron Aupperlee  Tuesday, Nov. 21, 2017, 11:06 a.m.


Argo AI, a Pittsburgh-based startup developing self-driving technology for Ford, will team up with researchers from Carnegie Mellon University and Georgia Tech to work on computer vision and machine learning technology.

Deva Ramanan and Simon Lucey from CMU and James Hays from Georgia Tech will join the Argo team, CEO Bryan Salesky wrote Tuesday in a post on Medium .

“These research scientists are playing an instrumental role in developing the core technologies that will allow self-driving cars both to see and understand the world around them, and to predict road user behavior,” Salesky wrote.

The researchers will remain on campus but spend time in Argo offices on a regular basis. They will continue to teach and maintain their work and research at their universities. Argo sees the collaboration as a way to link its work with academic research.

The researchers will also benefit from Argo funding set aside to support university research on self-driving cars, Salesky wrote.

Ford announced this year it will give Argo AI $1 billion over the next five years to develop self-driving technology for the automaker.

Ramanan and Lucey are at CMU’s Robotics Institute. Lucey leads the institute’s CI2CV Computer Vision Lab, where he trains computers to read facial expressions and body language. Ramanan’s work focuses on visual recognition and training computer programs to identify people by looking at body parts.

“Making sure self-driving cars can accurately identify people in all of their different shapes, sizes and positions is an essential step to establishing their safety and reliability,” Ramanan said in Salesky’s post.

Hays received his doctorate from CMU. At Georgia Tech, he uses data and crowdsourcing to help computers better understand scenes and images. He said working with Argo will allow him to design algorithms on a larger scale than in his previous research and use real-world data.

Aaron Aupperlee is a Tribune-Review staff writer. Reach him at aaupperlee@tribweb.com, 412-336-8448 or via Twitter @tinynotebook.