profile_picture
Sarah Bechtle
About me:
I am a last year PhD candidate at the Max Planck Institute for intelligent Systems in Tuebingen. I'm affiliated with the Computational Learning and Motor Control Lab, USC and the Machines in Motion Lab, NYU.

My research interests are at the intersection between machine learning and robotics - developing learning algorithms that can be deployed on robots. Specifically I'm interested in model based learning within the action-perception-learning loop of artificial agents with special interest in meta and lifelong learning.

I'm currently seeking research opportunities for projects associated with robotics, machine learning and computational neuroscience.

cvgoogle scholaremail

Recent news:
[March 2021] Our paper Leveraging Forward Model Prediction Error for Learning Control got accepted to ICRA

[January 2021] Our paper Meta-learning via learned loss won the overall best student paper award at ICPR

[January 2021] We are organizing a workshop on Learning to Learn for Robots at ICRA 2021

[December 2020] We open sourced our learning to learn code

[December 2020] We are organizing a workshop on Learning to Learn at ICLR 2021
Publications:
  1. Leveraging Forward Model Prediction Error for Learning Control
    Bechtle, S., Hammoud, B., Rai, A., Meier, F. and Righetti, L.
    IEEE International Conference on Robotics and Automation (ICRA), 2021

    [Paper][Video]

  2. Meta-learning via learned loss
    Bechtle, S., Molchanov, A., Chebotar, Y., Grefenstette, E., Righetti, L., Sukhatme, G. and Meier, F.
    IEEE International Conference on Pattern Recognition.

    Overall best student paper award

    [Paper][Video] [Code][Website]

  3. Model-Based Inverse Reinforcement Learning from Visual Demonstrations
    Das, N., Bechtle, S., Davchev, T., Jayaraman, D., Rai, A. and Meier, F.
    In Conference on Robot Learning, 2020

    [Paper][Video] [Code]

  4. Learning Extended Body Schemas from Visual Keypoints for Object Manipulation
    Bechtle, S., Das, N. and Meier, F.
    arXiv preprint, 2020

    [Paper][Video]

  5. Curious ilqr: Resolving uncertainty in model-based rl
    Bechtle, S., Lin, Y., Rai, A., Righetti, L. and Meier, F.
    In Conference on Robot Learning, 2019

    [Paper][Code and Video]