Multimodal Embodied Agents: A Unified Capability-centric Perspective from Computer-Use to Robot-Use
Resumo indisponível. Consulte o paper original.
PAPER / ARXIV:2609.15307
Yizhou Liu , Fei Tang , Yuchen Yan , Zhengxi Lu , Songqin Nong , Tao Jiang , Wenhao Xu , Wenqi Zhang , Weiming Lu , Jun Xiao , Yongliang Shen
RESUMO
Graphical User Interface (GUI) grounding is essential for autonomous agents to map natural language instructions to precise screen coordinates. However, existing supervised fine-tuning and reinforcement learning methods are constrained by the high cost of annotation, creating a scalability bottleneck. In this paper, we introduce a label-free test-time training paradigm driven by two key insights: (1) confidence patterns in coordinate tokens are a better indicator than full-sequence confidence, and (2) in sparse GUI coordinate spaces, negative samples offer more reliable learning signals than potentially noisy positive ones. We first propose Confidence-Anchored Learning (CAL), which utilizes coordinate-token confidence to filter pseudo-labels and assign distance-based binary rewards. Building on this, we develop Confidence-Anchored Negative Learning (CANL), which exclusively optimizes the model using negative samples to bypass the risks of incorrect positive samples. Experimental results demonstrate that CANL-7B achieves 92.1% on ScreenSpot-V2. On more challenging ScreenSpot-Pro, CANL-7B reaches 33.8%, an 8.9% absolute improvement over the base model. Our findings establish coordinate-token confidence as a powerful alternative to manual annotations for scalable GUI agent development.
NO MESMO MAPA
Resumo indisponível. Consulte o paper original.
Resumo indisponível. Consulte o paper original.
Resumo indisponível. Consulte o paper original.
Resumo indisponível. Consulte o paper original.
Resumo indisponível. Consulte o paper original.
Resumo indisponível. Consulte o paper original.