On Surprising Effects of Risk-Aware Domain Randomization for Contact-Rich Sampling-based Predictive Control
Sergio A. Esteban, Junheng Li, Vince Kurtz, Aaron D. Ames
TLDR
This paper explores risk-aware domain randomization in contact-rich predictive control, revealing its impact on both robustness and the cost landscape.
Key contributions
- First study of risk-aware domain randomization (DR) in contact-rich sampling-based predictive control (SPC).
- Compares average, optimistic, and pessimistic rollout aggregations on a simple Push-T task.
- Discovers DR not only improves robustness but also reshapes the cost landscape for contact actions.
Why it matters
This paper provides the first insights into how risk-aware domain randomization impacts contact-rich predictive control. It reveals that DR not only enhances robustness but also fundamentally alters the optimization landscape, guiding controllers towards better contact-producing actions. This understanding is crucial for developing more reliable and effective robotic systems in uncertain environments.
Original Abstract
Domain randomization (DR) is widely used in policy learning to improve robustness to modeling error, but remains underexplored in contact-rich sampling-based predictive control (SPC), where rollout quality is highly sensitive to uncertainty. In this work, we take the first step by studying risk-aware DR in predictive sampling on a simple yet representative Push-T task, comparing average, optimistic, and pessimistic rollout aggregations under randomized model instances. Our initial results suggest that DR affects not only robustness to model error, but also the effective cost landscape seen by the sampling-based optimizer, by reshaping the basin of attraction around contact-producing actions. This opens up potential for exploring better grounded risk-aware contact-rich SPC under model uncertainty. Video: https://youtu.be/f1F0ALXxhSM
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.