基于头眼行为的交互意图检测方法  

Identification method of interaction intention based on head and eye behaviors

作　　者：赵其杰[1,2] 邵辉[1] 卢建霞[1] 

机构地区：[1]上海大学机电工程与自动化学院,上海200072 [2]上海市智能制造及机器人重点实验室,上海200072

出　　处：《仪器仪表学报》2014年第10期2313-2320,共8页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金(61101177)资助项目

摘　　要：人与服务机器人的自然交互是服务机器人面向应用的重要问题,其中对交互意图的正确判断是提高交互自然性、可靠性的基础。提出了一种人—交互协作界面—移动机器人构成的交互系统,通过视觉检测头部和眼睛的动作行为,结合机器人端采集"观察"的场景,由头部确定某个兴趣区域,眼睛注视进一步从确定的区域精确化关注目标,进而确定当前的交互意愿。在建立的测试环境进行了实验,结果表明提出的方法可以有效地对用户意愿进行辨识,为深入研究人与服务机器人的合作探索了可行的新途径。Human-service robot natural interaction plays an important role for service robot application, and the correct judgment of inter- active intention is the foundation to improve interaction effect and reliability. This paper proposes a human-interactive collaboration inter- face-mobile service robot system; through detecting the head and eye behaviors based on machine vision approach, combining the feed- back of robot operating scene, the user＇ s head facing region is identified as the preliminary region of interest. Moreover, according to the direction of eye gaze the specific gazing target is determined; by this way, the user＇ s interested target can be followed and the inter- action intention can be identified. Finally, we established a test platform and some experiments were conducted; the results suggest that the proposed method can effectively identify the user＇ s interactive intention, which provides a new way for the further study of the collab- oration between service robot and human.

关 键 词：头眼行为 交互意图 协作交互界面 移动机器人 

分 类 号：TH39[机械工程—机械制造及自动化]