[논문 리뷰] AI Challenges in Human-Robot Cognitive Teaming

Tathagata Chakraborti, Subbarao Kambhampati외 2명|arXiv (Cornell University)|2017. 07. 15.

AI-based Problem Solving and Planning참고 문헌 45인용 수 61

한 줄 요약

이 논문은 인간의 인지 모델링이 인간-로봇 팀업의 효과에 필수적이라고 주장하며, HuM과 HuMM을 포함하는 업데이트된 에이전트 아키텍처를 제안하여 proactive, social, and explicable planning을 지원한다.

ABSTRACT

Among the many anticipated roles for robots in the future is that of being a human teammate. Aside from all the technological hurdles that have to be overcome with respect to hardware and control to make robots fit to work with humans, the added complication here is that humans have many conscious and subconscious expectations of their teammates - indeed, we argue that teaming is mostly a cognitive rather than physical coordination activity. This introduces new challenges for the AI and robotics community and requires fundamental changes to the traditional approach to the design of autonomy. With this in mind, we propose an update to the classical view of the intelligent agent architecture, highlighting the requirements for mental modeling of the human in the deliberative process of the autonomous agent. In this article, we outline briefly the recent efforts of ours, and others in the community, towards developing cognitive teammates along these guidelines.

연구 동기 및 목표

• Motivate why human cognitive expectations are central to human-robot teaming beyond physical coordination.
• Propose updating the classical agent architecture to include human models for deliberation and decision-making.
• Survey existing agent types and identify cognitive requirements for proactive and social behavior in robots.
• Outline challenges and approaches for human-aware planning, learning human models, communication, and evaluation in cognitive teaming.

제안 방법

• Expand the Sense-Model-Plan-Act (SMPA) cycle to include Human Model (HuM) and Human Mental Model (HuMM).
• Define C1-C3 capabilities: recognize teaming context, anticipate team behavior, and take actions that advance team goals while considering teammates.
• Discuss multi-model planning with HuM/HuMM, explicable planning, and model reconciliation explanations to align human and robot plans.
• Present learning and evaluation approaches for human models, including incomplete models and microworld testbeds for validation.

실험 결과

연구 질문

• RQ1 How should autonomous agents represent and reason about human teammates and their mental models (HuM/HuMM)?
• RQ2 What planning and explanation strategies are needed to produce human-aware, explicable, and trustworthy robot behavior?
• RQ3 How can robots learn, adapt, and communicate human models to improve long-term team performance?
• RQ4 What evaluation setups (microworlds) are suitable for studying human-robot cognitive teaming?
• RQ5 What are the key challenges in transitioning from traditional goal- and behavior-based agents to cognitive teaming agents?

주요 결과

• Cognitive teaming requires mental modeling of humans, not just physical/world states.
• An updated architecture with HuM and HuMM enables recognizing context, anticipating behavior, and acting to advance team goals while considering humans.
• explicable planning and model reconciliation can balance plan optimality with human understandability and trust.
• Learning incomplete human models and bi-directional communication are critical for long-term teaming performance.
• Microworlds and testbeds are valuable for rapid prototyping and validating human-robot cognitive teaming concepts.