Robots that work close to humans need to understand and use social cues to act in a socially acceptable manner. Social cues are a form of communication (i.e., information flow) between people. In this paper, a framework is introduced to detect and analyse social cues and information transfer directionality using an information-theoretic measure, namely, transfer entropy. We demonstrate the framework in three settings involving social interactions between humans: object-handover, group-joining and person-following. Results show that transfer entropy can identify information flows between agents, when and where they occur, and their relative strength. For instance, in a person-following scenario, we find that head orientation of a predictor is particularly informative, and the different times and locations that this is used to convey information to a leader influences their behaviour. Potential applications of the framework include information flow or social cue analysis for interactive robot design, or socially-aware robot planning.

翻译：与人类密切协作的机器人需要理解并运用社交线索，以符合社会规范的方式行动。社交线索是人际间的一种沟通形式（即信息流）。本文提出一种框架，利用信息论度量——传递熵，来检测和分析社交线索及信息传递方向性。我们在涉及人类社交互动的三种场景中验证该框架：物体交接、群体加入和人员跟随。结果表明，传递熵能够识别智能体之间的信息流、信息流发生的时空位置及其相对强度。例如，在人员跟随场景中，我们发现预测者的头部朝向携带特别丰富的信息，而该信息被用于向领导者传递意图的不同时间和位置会显著影响领导者的行为。该框架的潜在应用包括面向交互式机器人设计的信息流或社交线索分析，以及具有社会意识的机器人规划。