Jul 2026
CAC-VLA: Context-Gated Action Conditioning for Vision-Language-Action Models
CAC-VLA is proposed, a Context-Gated Action Conditioning framework that learns a lightweight latent-action interface directly within the VLM that enables VLM-native action conditioning while calibrating the influence of latent-action guidance on expert action generation.
Yi-Fu Xiong, Wen-Hao Yu, Jia-Xuan Lin et al.
· arXiv.org · 2 citations
· ⚡1