• Ph.D.
• Principal Scientist @ Autel Robotics
• Linkedin profile   Google Scholar profile

• I am currently at Autel Robotics working on Visual-Inertial SLAM. Previously, I was at Google AR during 2019-2021, where I have made two major contributions to ARCore: 36% decrease in CPU usage of motion tracking, and 28% reduction in motion tracking resets, both highlighted at Google I/O 2021.
• I received my PhD at Texas A&M University, where I developed a visual odometry system based on heterogeneous landmarks, as well as an RGB-D odometry algorithm solely based on line landmarks, being the first of its kind. Before joining Google, I worked at Honda Research, NVIDIA, and Amazon Lab126 on a varieaty of cool projects. At NVIDIA we developed a top-notch visual localization solution that showcased the possbility of lidar-free autonomous driving on highway. Check out the amazing demo videos!

• 2021-05-20: Two projects I led at Google AR are highlighted in Google I/O 2021.
  • I developed a feature that decreased the CPU usage of ARCore motion tracking (SLAM) by 36%.
  • I created a new algorithm that reduced ARCore tracking resets by 28%.
• 2020-11-28: Two U.S. patents are granted for my previous work on Autonomous Driving at Honda Research: "Sign based localization" and "Monocular localization in urban environments using road markings".
• 2018-02-20: A deep feature learning work by my summer intern (Kun He) is accepcted to CVPR 2018. Good job, Kun!
• 2017-10-09: I joined NVIDIA's Autonomous Vehicle Group.
• 2017-09-05: A paper on asynchronous visual-lidar SLAM is accepted to ISRR 2017.
• 2015-07-01: Two papers on RGBD odometry are accepted to IROS15 and ICCV15, respectively.

© Yan Lu |