Situation-aware proactive service by observing the user behaviors for service robots

Abstract

Robots are being used in industry for along time for the tasks that are hazardous and repetitive for human. Especially, robots are preferred in the industries for handling operations which require precision. Service robotics is a special kind of robotics technology that brings the smart robotic systems into our home. Human-robot interaction is considered to be a crucial aspect in designing Domestic service robots since they are regularly interacting with the human. In addition, the interaction between service robot and human should be smooth and natural. User behaviors are important aspect that should be considered in domestic service robot design to deliver an enhanced human-robot interaction. User behaviors include expression of emotions, preference, context and engagement of the user. User behavior observation is a valuable in providing appropriate services and responses to the user. This work presents the a proactive interaction manager used to enhance the humanrobot interaction. The proposed proactive interaction manager has the capability of perceiving the emotion of the user from vision and voice. It can assess the user engagement and context of the user based on the vocal responses of the user. The goal of the proactive interaction manager is to provide intelligent suggestions, context-aware responses and user support and caring based on the user behavior. Several experiments were conducted to validate that the user acceptability of the services provided by the proposed proactive interaction manager. Results of the experiments were analysed statistically to validate the claim of the hypothesis. The results of the experiments strongly agree that the proposed system is capable of achieving user satisfaction considering multiple attributes that represent the quality of interaction. Moreover, the results from the experiments were interpretable thereby the human can understand why the system came to a specific decision while there are other options for selection. The experiments were conducted with the use of Moratuwa Intelligent Robot platform which consists of a Pioneer 3DX mobile platform with a Cyton Gamma 300 manipulator attached to it, along with a Kinect sensor. Voice inputs from the user are obtained using the microphone array embedded within the Kinect sensor, while voice output is delivered through two stereo speakers integrated with the robot platform.