Use and usability of software verification methods to detect behaviour interference when teaching an assistive home companion robot: A proof-of-concept study

被引：4

作者：

Koay K.L. ^{[1
,5
]}

Webster M. ^{[2
]}

Dixon C. ^{[3
]}

Gainer P. ^{[4
]}

Syrdal D. ^{[1
,5
]}

Fisher M. ^{[3
]}

Dautenhahn K. ^{[6
]}

机构：

[1] Department of Computer Science, School of Physics, Engineering and Computer Science, University of Hertfordshire, College Lane, Hertfordshire, Hatfield

[2] School of Computer Science and Mathematics, Liverpool John Moores University, Liverpool

[3] Department of Computer Science, University of Manchester, Manchester

[4] Department of Computer Science, University of Liverpool, Liverpool

[5] Department of Computer Science, School of Physics, Engineering and Computer Science, University of Hertfordshire, College Lane, Hertfordshire, Hatfield

[6] Departments of Electrical and Computer Engineering and Systems Design Engineering, University of Waterloo, ON

来源：

Paladyn | 2021年 / 12卷 / 01期

基金：

英国工程与自然科学研究理事会;

关键词：

behaviour interference; companion robots; formal verification; human-robot interaction;

D O I：

10.1515/pjbr-2021-0028

中图分类号：

学科分类号：

摘要：

When studying the use of assistive robots in home environments, and especially how such robots can be personalised to meet the needs of the resident, key concerns are issues related to behaviour verification, behaviour interference and safety. Here, personalisation refers to the teaching of new robot behaviours by both technical and non-technical end users. In this article, we consider the issue of behaviour interference caused by situations where newly taught robot behaviours may affect or be affected by existing behaviours and thus, those behaviours will not or might not ever be executed. We focus in particular on how such situations can be detected and presented to the user. We describe the human-robot behaviour teaching system that we developed as well as the formal behaviour checking methods used. The online use of behaviour checking is demonstrated, based on static analysis of behaviours during the operation of the robot, and evaluated in a user study. We conducted a proof-of-concept human-robot interaction study with an autonomous, multi-purpose robot operating within a smart home environment. Twenty participants individually taught the robot behaviours according to instructions they were given, some of which caused interference with other behaviours. A mechanism for detecting behaviour interference provided feedback to participants and suggestions on how to resolve those conflicts. We assessed the participants' views on detected interference as reported by the behaviour teaching system. Results indicate that interference warnings given to participants during teaching provoked an understanding of the issue. We did not find a significant influence of participants' technical background. These results highlight a promising path towards verification and validation of assistive home companion robots that allow end-user personalisation. © 2021 Kheng Lee Koay et al., published by De Gruyter.

引用

页码：402 / 422

页数：20