الملخص
Egocentric video offers a scalable source of physical interaction experience, yet translating it into robot-executable knowledge and enabling continual adaptation remain challenging. We introduce Zeva-Ego, a unified framework that learns physical priors from human experience and evolves through robot interaction. An Action-Centric Encoder (ACE) converts egocentric visual transitions into action-centered supervision for VLA mid-training, while In-Context Causal Learning (ICCL) enables parameter-free adaptation from action-effect feedback at deployment. Scaling Ego data to 10K hours improves RoboTwin success from 63.8% to 75.3%, matching 2K hours of robot demonstrations (74.7%), corresponding to an empirical data ratio of roughly 4-5:1. With accumulated interaction experience, ICCL further improves success from 58% to 89% within four attempts without parameter updates. These results demonstrate a scalable path toward embodied intelligence that learns from human experience and continuously improves through its own interaction.
الكلمات المفتاحية
الموضوع
بيانات النشر
- المجلة
- غير متاح
- وصول مفتوح
- وصول مفتوح أخضر
اقتبس هذه المقالة
APA 7
Huang, B., Ding, X., Chen, F., Li, K., Sun, W., Wu, H., Liu, Y., & Cao, T. (2026). Zeva-Ego: Egocentric Mid-Training with In-Context Causal Learning for Robot Manipulation. https://omanscience.com/ar/articles/zeva-ego-egocentric-mid-training-with-in-context-causal-learning-for-robot-manipulation
MLA 9
Huang, Bingjia, et al. "Zeva-Ego: Egocentric Mid-Training with In-Context Causal Learning for Robot Manipulation." https://omanscience.com/ar/articles/zeva-ego-egocentric-mid-training-with-in-context-causal-learning-for-robot-manipulation.
شيكاغو (المؤلف–التاريخ)
Huang, Bingjia, Xin Ding, Fu Chen, Kun Li, Wei Sun, Hao Wu, Yunxin Liu, and Ting Cao. 2026. "Zeva-Ego: Egocentric Mid-Training with In-Context Causal Learning for Robot Manipulation." https://omanscience.com/ar/articles/zeva-ego-egocentric-mid-training-with-in-context-causal-learning-for-robot-manipulation.
هارفارد
Huang, B., Ding, X., Chen, F., Li, K., Sun, W., Wu, H., Liu, Y. and Cao, T. (2026) 'Zeva-Ego: Egocentric Mid-Training with In-Context Causal Learning for Robot Manipulation', Available at: https://omanscience.com/ar/articles/zeva-ego-egocentric-mid-training-with-in-context-causal-learning-for-robot-manipulation.
فانكوفر
Huang B, Ding X, Chen F, Li K, Sun W, Wu H, et al. Zeva-Ego: Egocentric Mid-Training with In-Context Causal Learning for Robot Manipulation. https://omanscience.com/ar/articles/zeva-ego-egocentric-mid-training-with-in-context-causal-learning-for-robot-manipulation
IEEE
B. Huang, X. Ding, F. Chen, K. Li, W. Sun, H. Wu, Y. Liu, and T. Cao, "Zeva-Ego: Egocentric Mid-Training with In-Context Causal Learning for Robot Manipulation," https://omanscience.com/ar/articles/zeva-ego-egocentric-mid-training-with-in-context-causal-learning-for-robot-manipulation.