Robust Odometry Estimation for RGB-D Cameras HD

This video demonstrates our robust visual odometry system using RGB-D data. Our system estimates the camera motion by minimizing the photometric error of consecutive RGB-D frames from a Microsoft Kinect / Asus Xtion Pro Live sensor. By pixel-wise weighting using the t-distribution, our solution is much more robust to dynamic objects in the scene and other sources of photometric noise. The software runs in real-time on a recent CPU. Publication: Robust Odometry Estimation for RGB-D Cameras (C. Kerl, J. Sturm, D. Cremers), In Proc. of the IEEE Int. Conf. on Robotics and Automation (ICRA), 2013 For more information and source code visit: http://vision.in.tum.de/data/software/dvo