UofT Robotics PhDs Anas Mahmoud (above left) and Juan Carrillo (above right) have been awarded the 2021 Qualcomm Innovation Fellowship (North America Program) for their proposal on sensor fusion for 3D object detection. Mahmoud and Carillo are one of 16 winning teams — and the only Canadian team — selected from among the 43 finalists.
Over 100 proposals are submitted each year to the Qualcomm Innovation Fellowship (QIF) program, which recognizes PhD students from across a range of technical research areas. Winning students earn a one year fellowship and are mentored by Qualcomm engineers to facilitate the success of the proposed research.
Mahmoud and Carillo’s winning proposal aims to develop novel deep learning models to enhance robot perception at long-range. Their proposal addresses the sparsity of range sensor measurements (i.e., LiDAR/RADAR) at long-range by fusing dense image data beyond direct pixel-point correspondences. They also introduce a 3D perception model that uses implicit representations of shape priors along with images to improve the characterisation of faraway objects. The application of their research will potentially allow autonomous vehicles and ADAS systems to operate safely at higher speeds by sensing obstacles at long-range, which enables planning and control algorithms to perform earlier decisions.
Mahmoud and Carillo are thrilled, and report they “are excited to tackle the challenge of detection at long range for its potential to enable various autonomous driving downstream tasks.” The pair added that their proposal “builds on previous object detection research at Prof. Waslander’s TRAILab, and the Qualcomm Innovation Fellowship will provide us with the resources needed to address this challenge.”
They are particularly looking forward to the industry partnerships the fellowship creates. “The fellowship offers mentorship from senior staff at Qualcomm, and funding support to two Ph.D. students amounting to 100K USD for a year,” they reported.
Mahmoud and Carillo are members of Prof. Steve Waslander’s TRAIL Lab, which focuses on robotic perception and planning problems in the domains of unmanned aerial vehicles and self-driving cars. Steve Waslander is a recent recipient of a $12M ORF-RE in All-Weather Autonomous Driving.