Reproduce “Good Time to Ask: A Learning Framework for Asking for Help in Embodied Visual Navigation” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jenny Zhang, Samson Yu, Jiafei Duan, Cheston Tan
The paper introduces a learning framework called "Good Time to Ask" (GTA) that enables embodied agents to actively request semantic segmentation feedback from a human teacher during Object-Goal Navigation tasks. This approach improved navigation success rates by an average of 31.7% and enhanced robustness to intermittent teacher availability through a semi-present teacher training curriculum.
Key insights
- Actively seeking ground truth semantic segmentation feedback from a teacher substantially enhances an embodied agent's performance and efficiency in Object-Goal Navigation tasks.
- A semi-present teacher training curriculum is crucial for preventing agent over-reliance on continuous feedback, ensuring robust performance even when assistance is intermittently available.
- The effectiveness of "ask" actions can be quantitatively assessed by their immediate impact on reducing the agent's uncertainty regarding the target object's location, allowing for optimized query strategies.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.