Peretroukhin with his thesis committee (from left to right: Angela Schoellig, Animesh Garg, Valentin Peretroukhin, Jonathan Kelly, and Tim Barfoot) in early March.
STARS Lab graduate Valentin Peretroukhin was recognized with the Gordon N. Patterson award for the top PhD thesis at UTIAS this year for his dissertation titled Learned Improvements to the Visual Egomotion Pipeline.
“I am thrilled and honoured to receive the Gordon N. Patterson award,” Peretroukhin says. “I am immensely grateful to my supervisor, Prof. Jonathan Kelly, for his support, guidance, and friendship, and to my lab mates and co-conspirators without which this work would not have been possible.”
Peretroukin’s thesis explores visual egomotion estimation, also known as visual odometry, which is the process whereby a robot estimates its position and orientation using only the images captured from its onboard cameras. As the cost of high-quality compact cameras has gone down, visual odometry has emerged as an important method for robots to localize themselves as they explore the world. With the goal of enabling safe long-term autonomy, Peretroukhin’s thesis presents four ways to augment classical approaches to visual odometry with machine learning, to make them more robust and accurate in complex environments. Peretroukhin is currently a Postdoctoral Researcher in the Robust Robotics Group with Prof. Nicholas Roy at MIT CSAIL.
Congratulations to Valentin Peretroukhin from the UofT Robotics Institute!