Research Highlights

Deep Probabilistic Rotation Estimation

Using a multi-head structure, we build a deep network that can regress uncertain estimates of rotation using quaternion averaging and noise injection. This work is currently published in the CVPR Workshop on Uncertainty in Visual Learning as a workshop paper.

DPC-Net: Deep Pose Correction

Deep Pose Correction Network (DPC-Net) is a deep learning approach to achieving accurate, robust visual localization by correcting existing geometric, probabilistic estimators. It's been published in Robotics and Automation Letters (and presented at ICRA 2018).

Sun-BCNN

Sun-BCNN (Bayesian Convolutional Neural Network) infers the full 3D sun direction from single RGB images (which can then be used to improve localization).

PROBE: Predictive Robust Estimation

PROBE, and its follow-up, PROBE-GK (Generalized Kernels) is a heteroscedastic covariance model for image keypoints in visual odometry. Notably, PROBE-GK presents a novel Bayesian formulation for deriving a robust cost function for sparse keypoint reprojection errors.


Recent accomplishments

May, 2019 - After months of planning, my colleagues Lee, Matt and supervisor Jonathan Kelly organized a debates workshop at ICRA 2019. We were elated to see a fantastic turn-out for all three debates (plus lightning talks) and a ton of great feedback!

Post-Solo Shower

Final Approach

April, 2018 - After months of bad weather and cancelled lessons, I can finally say I’m a private pilot! I passed my flight test in late March. My girlfriend, Elyse, was my first passenger.

August, 2017 - I’ve been working on my private pilot license (PPL) this summer at Burlington Executive Airport (CZBA). After two months of training, I finished my first solo flight (which means I was the only person in the aircraft). If you’re interested in seeing the whole flight, I recorded the whole thing on my GoPro .

Post-Solo Shower

Final Approach