ShieldVLA: Feasibility-Aware Safety Alignment for Vision-Language-Action Models
ShieldVLA, a safety-aligned fine-tuning framework for VLA models, improves safety guarantees via Hamilton-Jacobi reachability.
A study proposes ShieldVLA, a safety-aligned fine-tuning framework for Vision-Language-Action (VLA) models, using Hamilton-Jacobi reachability to improve safety guarantees.
- Proposes a safety-aligned fine-tuning framework for VLA models
- Reduces cumulative safety cost by 57% on benchmarks
- Improves task success rate by 0.13 over SafeVLA
- Employs HJ reachability to estimate safe operating regions
- Countries
- US
From the original article
1 min read at huggingface.co
Vision-Language-Action (VLA) models demonstrate strong generalization in robotic manipulation and navigation, but existing fine-tuning methods provide limited safety guarantees. Current approaches primarily rely on Lagrangian optimization that enforces safety through soft penalties on expected cumulative cost, often resulting in residual constraint violations or overly conservative behavior.
By Manan Tayal, Akshay Nambi · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.