Michael Baumgartner
Michael Baumgartner is a PhD student jointly affiliated with ETH Zurich and Disney Research, where his work spans computer vision, robotics, deep learning, and mixed reality applications. He is best known as the lead author of CoCo-InEKF, a paper presenting a differentiable invariant extended Kalman filter that uses learned continuous contact covariances for state estimation in dynamic, contact-rich legged robot scenarios, presented at Robotics: Science and Systems. His broader research interests include machine learning applied to robots as edge devices and robot-human interaction.
“CoCo-InEKF: State Estimation with Learned Contact Covariances in Dynamic, Contact-Rich Scenarios”
Source→“We are eager to explore whether incorporating real-world data or greater training diversity can further improve performance.”
Source→AI-extracted from podcast / newsletter / paper summaries. May contain errors.