Predicting the Intention to Interact with a Service Robot:the Role of Gaze Cues

Simone Arreghini,Gabriele Abbate,Alessandro Giusti,Antonio Paolillo

2024-04-02

Abstract:For a service robot, it is crucial to perceive as early as possible that an approaching person intends to interact: in this case, it can proactively enact friendly behaviors that lead to an improved user experience. We solve this perception task with a sequence-to-sequence classifier of a potential user intention to interact, which can be trained in a self-supervised way. Our main contribution is a study of the benefit of features representing the person's gaze in this context. Extensive experiments on a novel dataset show that the inclusion of gaze cues significantly improves the classifier performance (AUROC increases from 84.5% to 91.2%); the distance at which an accurate classification can be achieved improves from 2.4 m to 3.2 m. We also quantify the system's ability to adapt to new environments without external supervision. Qualitative experiments show practical applications with a waiter robot.

Robotics,Artificial Intelligence,Machine Learning

What problem does this paper attempt to address?

### Problems the Paper Aims to Solve

This paper aims to address the issue of how service robots can anticipate human interaction intentions in advance. Specifically:

1. **Predicting Interaction Intentions**: The study proposes a method that enables service robots to predict the interaction intentions of any person within their field of view through self-supervised learning. This task is crucial for enhancing user experience, as robots can proactively exhibit friendly behavior when users approach, thereby improving the user experience.

2. **Importance of Non-verbal Communication**: The research emphasizes the importance of non-verbal communication (such as eye contact) in human-robot interaction and validates through experiments that incorporating eye contact cues significantly improves the classifier's performance. Experimental results show that after adding eye contact cues, the classifier's AUROC increased from 84.5% to 91.2%, and it could make accurate predictions earlier (distance increased from 2.4 meters to 3.2 meters).

3. **Adaptive Capability**: The study also explores the system's ability to adapt to new environments and demonstrates the friendly behavior of service robots in practical applications.

In summary, the main goal of this paper is to improve service robots' ability to predict potential user interaction intentions by introducing eye contact cues and to demonstrate the effectiveness of this approach through experiments.