Quartz 5

How Do We Research Human-Robot Interaction in the Age of LLMs? A Systematic Review

Properties2
descriptionSystematic review of 86 articles on LLM-driven HRI, identifying how LLMs reshape contextual sensing, social interaction, and alignment with human needs in physical environments.
tagsclippings

1 min read

How Do We Research HRI in the Age of Large Language Models?

Authors: Yufeng Wang, Yuan Xu, Anastasia Nikolova, Yuxuan Wang, Jianyu Wang, Chongyang Wang, Xin Tong Venue: CHI 2026 (ACM Conference on Human Factors in Computing Systems)

Overview

A systematic literature review of 86 articles examining how Large Language Models are transforming Human-Robot Interaction. Follows PRISMA guidelines. Focus on human-centered perspectives that existing purely technical reviews overlook.

Key Contributions

  • Systematic examination of LLMs’ human-centered impact: human understanding, user modeling, and levels of autonomy
  • Consolidated overview of emerging challenges in LLM-driven HRI
  • Design considerations and future research guidelines

Key Findings

  • LLMs fundamentally reshape how robots perceive context and generate socially appropriate responses
  • LLMs enable contextual sensing: robots can infer situational context from language rather than relying solely on sensor data
  • LLMs enable socially grounded interactions: natural dialogue, intent inference, common-sense reasoning
  • Current research is exploratory with inconsistent experimental setups and evaluation metrics
  • Gap: human-centered considerations (user modeling, appropriate autonomy levels) are underrepresented vs. technical implementation papers

Research Gaps Identified

  • Lack of standardized evaluation methodology for LLM-driven HRI systems
  • Insufficient attention to user modeling and personalization
  • Levels of autonomy rarely explicitly designed or studied
  • Human-oriented understanding lags behind technical advances

Graph View