Intelligent construction is an important construction trend. With the growing number of intelligent autonomous systems implemented in the construction area, understanding and predicting human motion becomes increasingly important. Based on such predictions, the autonomous systems can optimize their actions to improve the efficiency of human-robot interactions, and supervisors can make informed decisions about when and where to intervene in human motion to avoid collisions. This paper presents a comprehensive review of existing literature on human motion prediction (HMP). Relevant studies from a wide range of fields are reviewed, analyzed and synthesized, in terms of prediction indicators, methods and applications, based on a three-level taxonomy. The taxonomy is structured based on the levels of human information required by different prediction methods, and reflects different understandings of the underlying causality and mediators of human motions and intent. The paper also discusses the evolutions of the theoretical understanding and methodological development of HMP, its application scenarios in and beyond the construction domain, and possible directions for future research. This review is expected to increase the visibility of this rapidly expanding research area, and inspire future studies and advancements for human-robot interactions in construction.

智能化建设是重要的建设趋势。随着越来越多的智能自主系统在建筑领域实施，理解和预测人体运动变得越来越重要。基于这样的预测，自主系统可以优化它们的动作以提高人机交互的效率，并且监督者可以就何时何地干预人体运动以避免碰撞做出明智的决定。本文全面回顾了有关人体运动预测 (HMP) 的现有文献。在三级分类的基础上，从预测指标、方法和应用等方面对来自广泛领域的相关研究进行回顾、分析和综合。该分类是根据不同预测方法所需的人类信息水平构建的，反映了对人类运动和意图的潜在因果关系和中介的不同理解。本文还讨论了 HMP 理论理解和方法发展的演变、其在建筑领域内外的应用场景以及未来研究的可能方向。这篇综述有望提高这个快速扩展的研究领域的知名度，并激发未来在建筑中人机交互的研究和进步。其在建筑领域内外的应用场景，以及未来可能的研究方向。这篇综述有望提高这个快速扩展的研究领域的知名度，并激发未来在建筑中人机交互的研究和进步。其在建筑领域内外的应用场景，以及未来可能的研究方向。这篇综述有望提高这个快速扩展的研究领域的知名度，并激发未来在建筑中人机交互的研究和进步。