This paper is dedicated to F-2 the interlocutor robot which is a platform for approbating strategies and elements of multimodal communicative behavior. Project F-2 follows the trail blazed by Marvin Minsky: a robot is controlled by interchanging communicative units produced as response to unified representation of heterogeneous input stimuli. We describe the artificial interlocutor architecture and further revision of its controller which processes the activated communicative goals pool. It is on the basis of this unit that personal traits are to be introduced. We overview existing factors which are to be considered in order to define context via modifiers of conducting dialogue. Next, with regard to developing personal traits we introduce temperament into our model. We describe data collected on the preferred choice of different response scenarios in specific situations by people of different temperaments; we implement activated scenarios re-weighing. For approbation we survey respondents on the plausibility, adequacy and pleasantness of the modified robot behaviour conditioned by a given temperament, as compared to basic reactions. According to assessors, personal traits are successfully transferred into the robot’s behaviour in the selected set of situations, which is pointed out as a proof of concept.

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Towards Developing Personal Traits of F-2 the Communicative Robot: Context and Temperament

  • Liliya Volkova,
  • Artemy Kotov,
  • Kirill Kivva,
  • Eseniya Vinogradova

摘要

This paper is dedicated to F-2 the interlocutor robot which is a platform for approbating strategies and elements of multimodal communicative behavior. Project F-2 follows the trail blazed by Marvin Minsky: a robot is controlled by interchanging communicative units produced as response to unified representation of heterogeneous input stimuli. We describe the artificial interlocutor architecture and further revision of its controller which processes the activated communicative goals pool. It is on the basis of this unit that personal traits are to be introduced. We overview existing factors which are to be considered in order to define context via modifiers of conducting dialogue. Next, with regard to developing personal traits we introduce temperament into our model. We describe data collected on the preferred choice of different response scenarios in specific situations by people of different temperaments; we implement activated scenarios re-weighing. For approbation we survey respondents on the plausibility, adequacy and pleasantness of the modified robot behaviour conditioned by a given temperament, as compared to basic reactions. According to assessors, personal traits are successfully transferred into the robot’s behaviour in the selected set of situations, which is pointed out as a proof of concept.