Embodied Vision Technical Report 2024

Incremental Few-Shot Adaptation for Non-Prehensile Object Manipulation using Parallelizable Physics Simulators

Few-shot adaptation is an important capability for intelligent robots that perform tasks in open-world settings such as everyday environments or flexible production. In this paper, we propose a novel approach for non-prehensile manipulation which iteratively adapts a physics-based dynamics model for model-predictive control. We adapt the parameters of the model incrementally with a few examples of robot-object interactions. This is achieved by sampling-based optimization of the parameters using a parallelizable rigid-body physics simulation as dynamic world model. In turn, the optimized dynamics model can be used for model-predictive control using efficient sampling-based optimization. We evaluate our few-shot adaptation approach in several object pushing experiments in simulation and with a real robot.

Author(s): Baumeister, Fabian and Mack, Lukas and Stueckler, Joerg
Book Title: CoRR abs/2409.13228
Year: 2024
Bibtex Type: Technical Report (techreport)
Electronic Archiving: grant_archive
Institution: CoRR
Note: Submitted to IEEE International Conference on Robotics and Automation (ICRA) 2025
State: Submitted
URL: https://arxiv.org/abs/2409.13228

BibTex

@techreport{baumeister2024mpcsim,
  title = {Incremental Few-Shot Adaptation for Non-Prehensile Object Manipulation using Parallelizable Physics Simulators},
  booktitle = {CoRR abs/2409.13228},
  abstract = {Few-shot adaptation is an important capability for intelligent robots that perform tasks in open-world settings such as everyday environments or flexible production. In this paper, we propose a novel approach for non-prehensile manipulation which iteratively adapts a physics-based dynamics model for model-predictive control. We adapt the parameters of the model incrementally with a few examples of robot-object interactions. This is achieved by sampling-based optimization of the parameters using a parallelizable rigid-body physics simulation as dynamic world model. In turn, the optimized dynamics model can be used for model-predictive control using efficient sampling-based optimization. We evaluate our few-shot adaptation approach in several object pushing experiments in simulation and with a real robot. },
  institution = {CoRR},
  year = {2024},
  note = {Submitted to IEEE International Conference on Robotics and Automation (ICRA) 2025},
  slug = {baumeister2024mpcsim},
  author = {Baumeister, Fabian and Mack, Lukas and Stueckler, Joerg},
  url = {https://arxiv.org/abs/2409.13228}
}