ICON Seminar in Autonomy/Physical AI: Dr. Yin Zhou (Waymo)

Time: 3:00-4:00 pm Eastern Time, February 23 (Friday), 2024

Location: PGSC 105 (Purdue Graduate Student Center) PGSC is the office building in front of the Northwestern parking garage.

Zoom Link: https://purdue-edu.zoom.us/j/98798335169?pwd=ZFNMdmZXendoQ1RCRDczM2dTR1RIdz09

Free coffee and snacks will be provided.     

Total Scene Understanding for Autonomous Driving

Abstract:

Within the autonomy stack, model training typically requires a large amount of labels. Manual annotation is expensive in terms of time and money. To reduce cost and improve efficiency, it is important to automate the labeling process. In this talk, I will introduce our auto labeling technologies, which can achieve on-par quality as experienced annotators and thus enable us to more efficiently understand the semantics of the world. The real world, however, usually contains long-tail instances and unseen categories that may not be well represented by the training data. How to handle these cases poses new challenges to existing autonomy systems. I will also share our pioneering works on open-set perception and prediction, which are critical capabilities for autonomous vehicles to handle the unknowns when navigating through the complex world.

Speaker:

Dr. Yin Zhou currently works as a senior staff research scientist and senior manager at Waymo. His team has developed cutting-edge technologies for onboard perception, prediction, and planning as well as offboard auto labeling and simulation. Prior to Waymo, he was a tech lead in Apple Special Project Group and a senior research scientist at Samsung Research America. He obtained his Ph.D. degree in Electrical Engineering from the University of Delaware in 2014 and his Bachelor's degree in Communications Engineering from Beijing Jiaotong University in 2009, respectively. 

Recording:

https://www.youtube.com/watch?v=7mU5hvV0HpI&ab_channel=ICONPurdue

Organizers: Ziran Wang (ziran@purdue.edu), Nak-seung Patrick Hyun (nhyun@purdue.edu), & Tho Le (thovle@purdue.edu)