The authors model robot behavior using the joint trajectory distribution induced by a frozen vision-language-action policy and the environment.
Looking Beyond Safe Actions Helps Robots Avoid Policy Dead Ends
A training-free decoder evaluates whether candidate robot actions preserve policy-supported paths to safe task completion.
Chinese Tech
Tu Nguyen · Matthieu Zimmer · Vu Anh Vu · Ziyi Wang · Jannik Hammel Nielsen · Xuebing Zhou · +1 more
Huawei Heisenberg Research Center · Huawei Noah’s Ark Lab · TU Berlin · UCL Center for AI
Research Digest··3 min read
Nguyen et al.
Why this paper
From Huawei Heisenberg Research Center and 3 others
In one line
Safe actions can lead to dead ends; this work reranks action candidates based on the safe completions they leave open.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors (2 noted)
- ✓Reports numbers on named benchmarks
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§