11 papers
Zeva-Ego: Egocentric Mid-Training with In-Context Causal Learning for Robot Manipulation
Bingjia Huang, Xin Ding, Fu Chen +5
Egocentric video offers a scalable source of physical interaction experience, yet translating it into robot-executable knowledge and enabling continual adaptation remain challengin…
Zeva: In-Context Causal Learning for Generalizable Embodied Manipulation
Fu Chen, Xin Ding, Bingjia Huang +8
Generalizable embodied manipulation remains difficult to achieve through pretraining alone, due to unseen physical conditions in the real world. We argue that robots need to learn…
Zetta : An Efficient Closed-Loop Embodied Harness for Self-Evolving Physical Intelligence
Xin Ding, Liang Mi, Mingzhe Huang +12
Embodied agents are increasingly used to close the gap left by end-to-end policy models. Yet the agentic path has not realized closed-loop learning in physical execution: existing…
GRIP-VLM: Group-Relative Importance Pruning for Efficient Vision-Language Models
Mingzhe Huang, Weijun Wang, Xin Ding +7
In Vision-Language Models (VLMs), processing a massive number of visual tokens incurs prohibitive computational overhead. While recent training-aware pruning methods attempt to sel…
EmbodiSkill: Skill-Aware Reflection for Self-Evolving Embodied Agents
Ruofei Ju, Xinrui Wang, Xin Ding +12
Embodied agents can benefit from skills that guide object search, action execution, and state changes across diverse environments. Since embodied environments vary across layouts,…
Em-Garde: A Propose-Match Framework for Proactive Streaming Video Understanding
Yikai Zheng, Xin Ding, Yifan Yang +6
Recent advances in Streaming Video Understanding has enabled a new interaction paradigm where models respond proactively to user queries. Current proactive VideoLLMs rely on per-fr…
OxyGen: Unified KV Cache Management for VLA Inference under Multi-Task Parallelism
Xiangyu Li, Huaizhi Tang, Xin Ding +3
Embodied AI agents increasingly require parallel execution of multiple tasks, such as manipulation, conversation, and memory construction, from shared observations under distinct t…
ROCKET: Residual-Oriented Multi-Layer Alignment for Spatially-Aware Vision-Language-Action Models
Guoheng Sun, Tingting Du, Kaixi Feng +6
Vision-Language-Action (VLA) models enable instruction-following robotic manipulation, but they are typically pretrained on 2D data and lack 3D spatial understanding. An effective…
AdaNav: Adaptive Reasoning with Uncertainty for Vision-Language Navigation
Xin Ding, Jianyu Wei, Yifan Yang +10
Vision Language Navigation (VLN) requires agents to follow natural language instructions by grounding them in sequential visual observations over long horizons. Explicit reasoning…
StreamMind: Unlocking Full Frame Rate Streaming Video Dialogue through Event-Gated Cognition
Xin Ding, Hao Wu, Yifan Yang +4
With the rise of real-world human-AI interaction applications, such as AI assistants, the need for Streaming Video Dialogue is critical. To address this need, we introduce StreamMi…
Making Every Frame Matter: Continuous Activity Recognition in Streaming Video via Adaptive Video Context Modeling
Hao Wu, Donglin Bai, Shiqi Jiang +6
Video activity recognition has become increasingly important in robots and embodied AI. Recognizing continuous video activities poses considerable challenges due to the fast expans…