Space and Terrestrial Autonomous Robotic Systems Lab at the University of Toronto
Pixels to pixels – RA-L and IROS 2020 paper on learning latent dynamics for control from pixels!
New RA-L and IROS 2020 work with the RLAI Lab at the University of Alberta on learning robust latent dynamics from images only. We demonstrate how latent dynamics can be made more robust for real-world robotic applications by composing a Kalman filter with a learned, heteroskedastic uncertainty model. Watch the YouTube video to find out more!
