The use of emotion to drive robotic interaction continues to grow across a range of use cases, from social robotics to increased survivability. Nevertheless, these efforts remain isolated from each other and are not easily compared between papers and projects. To this end an extensive survey of 1427 IEEE and ACM publications was conducted, covering robotics and emotion. The survey first resulted in broad categorizations of key trends covering emotional input and output. This was followed by an extended analysis on 232 papers that focused on the internal processing of emotion, where emotion was handled through some kind of algorithm and not just as an input or output. From this analysis, three broad categories were developed: emotional intelligence, emotional model, and implementation. Emotional intelligence captured the manner in which emotion was handled and included the subcategories: algorithm, mapping, and history. The emotional model category captured the emotion categories and number of emotions used, while the implementation category tracked the role, purpose, and platform. This paper concludes with a summary of key features discovered through the process, future opportunities, and a discussion of the intrinsic challenges emerging from the interaction of emotion and robotics.

使用情感来驱动机器人交互在一系列用例中继续增长，从社交机器人到提高生存能力。尽管如此，这些努力仍然彼此孤立，并且不容易在论文和项目之间进行比较。为此，对 1427 份 IEEE 和 ACM 出版物进行了广泛调查，涵盖机器人技术和情感。该调查首先对涵盖情绪输入和输出的关键趋势进行了广泛的分类。随后对 232 篇论文进行了扩展分析，这些论文侧重于情绪的内部处理，其中情绪是通过某种算法处理的，而不仅仅是作为输入或输出。根据这一分析，开发了三大类：情商、情感模型和实施。情商捕捉了处理情绪的方式，包括以下子类别：算法、映射和历史。情感模型类别捕获情感类别和使用的情感数量，而实施类别跟踪角色、目的和平台。本文最后总结了在这个过程中发现的关键特征、未来的机会，以及对情感和机器人交互中出现的内在挑战的讨论。