I presented two papers at the CoRoboLearn: Advancing Learning for Human-Centered Collaborative Robots workshop at the Conference on Robot Learning (CoRL) 2024 in Munich, Germany.
Our first-author paper on interactive incremental skill learning was selected for a spotlight presentation:
Spotlight: Interactive Incremental Imitation Learning of Generalizable Skills with Local Trajectory Modulation Link to heading
Knauer, M., Albu-Schäffer, A., Stulp, F., Silvério, J.
We present a method for interactive, incremental learning of generalizable robot manipulation skills using task-parameterized kernelized movement primitives. Physical corrections transfer across task configurations through local trajectory modulation, and uncertainty-driven variable impedance ensures safe interaction.
Co-authored: Human-Intention-Aware Skill Modulation Using Energy Tanks for Collaborative Tasks Link to heading
Fiorini, E., Knauer, M., Silvério, J.
An energy-tank framework that identifies human intentions during physical interaction across three Cartesian degrees of freedom, dynamically adjusting impedance control for intuitive skill corrections.