Topological mapping based on perceiving-acting cycle in sharing cognitive environments for robot partners

Fernando Ardilla (Graduate School of System Design, Tokyo Metropolitan University, Hino, Tokyo, Japan; and Dept. of Information and Computer Engineering, Politeknik Elektronika Negeri Surabaya, Indonesia)

Mohamad Yani (Graduate School of System Design, Tokyo Metropolitan University, Hino, Tokyo, Japan; and Dept. of Computer Engineering, Faculty of Electrical Engineering, Institut Teknologi Telkom Surabaya, Indonesia)

Azhar Aulia Saputra (Graduate School of System Design, Tokyo Metropolitan University, Hino, Tokyo, Japan)

Weihong Chin (Graduate School of System Design, Tokyo Metropolitan University, Hino, Tokyo, Japan)

Naoyuki Kubota (Graduate School of System Design, Tokyo Metropolitan University, Hino, Tokyo, Japan)

Abstract

Various kinds of human-friendly robot partners have recently been developed to provide humans with superior services. Manipulation skills, including grasping, arranging, and delivering, are essential for home applications. A robot partner is designed to grasp the meaning of human behavior and their intention in shared spaces to assist older people at home. As a result, the robot partner requires the cognitive ability to comprehend states of the environment based on both people’s and robot partners’ physical and sensory embodiment. This research presents a human-robot interaction technique for handover behaviors based on cognitive contexts. First, we describe how to share a person’s cognitive environment with a robot companion using the relevance idea presented in Cognitive Pragmatics. The perceived cognitive environment of humans contains a type of spatial topological structure, such as relative placement and proximity among objects. Furthermore, the human cognitive environment is continually updated due to the cyclic process of perception and action. As a result, we will look at how to apply topological mapping approaches in cognitive contexts. Next, using the idea of the perceiving-acting cycle presented in Ecological Psychology, we apply topological mapping methods of Growing Cell Structure (GCS) and Growing Neural Gas (GNG). The GCS represents the effectivity in the action system. In contrast, the GNG represents the human and robot task space. The experimental findings and real-world robot application examples indicate that the robot can correctly estimate human intention and conduct handover actions. Finally, we examine the effectiveness of the proposed approach and future research directions in the human-robot interaction based on the perceiving-acting cycle.

The full text of this article is unavailable through your IP address: 3.145.12.233

This work was partially supported by JST [Moonshot RnD][Grant NumberJPMJMS2034]

Received 13 January 2021

Published 18 January 2023