Summary of Human-agent Coordination in Games Under Incomplete Information Via Multi-step Intent, by Shenghui Chen et al.
Human-Agent Coordination in Games under Incomplete Information via Multi-Step Intent
by Shenghui Chen, Ruihan Zhao, Sandeep Chinchali, Ufuk Topcu
First submitted to arxiv on: 23 Oct 2024
Categories
- Main: Artificial Intelligence (cs.AI)
- Secondary: Human-Computer Interaction (cs.HC)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper extends a turn-based cooperative game framework to enable agents to take multiple actions per turn, allowing for multi-step intent modeling. This extension is hypothesized to improve performance in long-horizon tasks. To develop cooperative policies for the agent, the authors propose IntentMCTS, an online planning algorithm that considers the current belief and leverages communicated multi-step intent via reward augmentation. Agent-to-agent simulations demonstrate IntentMCTS requires fewer steps and control switches than baseline methods. A human-agent user study shows a higher success rate (18.52%) and improved cognitive load, frustration, and satisfaction compared to heuristic baselines. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Researchers developed a new way for machines and humans to work together more effectively. They created a game-like framework that lets machines make multiple decisions at once, which can help with long-term tasks. The team designed an algorithm called IntentMCTS that helps machines learn from each other and adjust their actions based on what they’ve learned. Simulations showed that this approach worked better than previous methods. In a test with humans, people working with the new algorithm had higher success rates and felt less stressed. |
