الباحثون

Duo Wu

المنشورات 2

نسخة أولية وصول مفتوح

Seek Before You Move: Evidence Seeking for Progress Grounding in Vision-Language Navigation

Zhimin Wang, Meiyuan Zhu, Duo Wu وآخرون · 2026

Vision-Language Navigation (VLN) requires agents to continuously ground task progress from long-horizon instructions and partial egocentric observations. Existing VLM-based navigation agents typically reason only over available observations and may remain confident even when task-relevant evidence is missing. For examp …

نسخة أولية وصول مفتوح

Demonstration-Free Success-Probability Reward Learning for Generalist Robot Policies

Duo Wu, Haifeng Wang, Rongwei Lu وآخرون · 2026

Reinforcement learning (RL) enables generalist robot policies to improve through trial-and-error interaction, yet its effectiveness is fundamentally constrained by sparse task rewards. Existing general-purpose reward models typically alleviate this issue by learning task progress from expert demonstrations, but introdu …

المؤلفون المشاركون