Show me your moves! Conveying navigation intention of a mobile robot to humans

Abstract

When humans and mobile robots share the same space, one of their challenges is to navigate around each other and manage their mutual navigational intents. While humans have developed excellent skills in inferring their counterpart's intentions via a number of implicit and non-verbal cues, making navigation also in crowds an ease, this kind of effective and efficient communication often falls short in human-robot encounters. In this paper, two alternative approaches to convey navigational intent of a mobile robot to humans in a shared environment are proposed and analysed. The first is utilising anthropomorphic features of the mobile robot to realise an implicit joint attention using gaze to represent the direction of navigational intent. In the second approach, a more technical design adopting the semantics of car's turn indicators, has been implemented. The paper compares both approaches with each other and against a control behaviour without any communication of intent. Both approaches show statistically significant differences in comparison to the control behaviour. However, the second approach using indicators has shown as being more effective in conveying the intent and also has a higher positive impact on the comfort of the humans encountering the robot.