Integrating accountability in the systems design of autonomous and remote-controlled operations

Bård Myhre,Ørnulf Jan Rødseth,Stig Petersen

doi:10.1088/1757-899x/929/1/012020

Bård Myhre, Ørnulf Jan Rødseth + Show 1 more

Open Access

https://doi.org/10.1088/1757-899x/929/1/012020

Copy DOI

Abstract

In a previous paper we have suggested that the transferal of human accountability from an on-site human actor (such as the captain) to a remote human actor (such as the creator of the autonomous control system) could be regarded as the defining characteristic of autonomous systems. In this paper we take this approach one step further, by suggesting a methodology for how accountability can be used as a basis for systems design of autonomous and remote-controlled operations. Furthermore, the suggested methodology is applied on a hypothetical case of a vessel supporting both autonomous and remote-controlled operation.

Highlights

The concept of autonomy is by many considered to be the paradigm shift within transportation
In a previous paper we have suggested that the transferal of human accountability from an on-site human actor to a remote human actor could be regarded as the defining characteristic of autonomous systems
Accountability as a starting point for autonomous and remote-controlled operations As mentioned in section 1, our previous paper [3] suggested a definition for autonomy that have been modified as follows: An automated system is considered autonomous within a specified operational envelope if it can lawfully accept accountability for an operation, thereby assuming the accountability that was previously held by either a human operator or another autonomous system

Summary

Introduction

The concept of autonomy is by many considered to be the paradigm shift within transportation. 2. Accountability as a starting point for autonomous and remote-controlled operations As mentioned, our previous paper [3] suggested a definition for autonomy that have been modified as follows: An automated system is considered autonomous within a specified operational envelope if it can lawfully accept accountability for an operation, thereby assuming the accountability that was previously held by either a human operator or another autonomous system. Note: As an automated system itself cannot be held legally accountable, the factual accountability for the actions of an autonomous system will necessarily reside with, e.g. the creators, installers or operators of the system, dependent on how the liability regime develops [7] This definition implies that a human operator most likely can be considered to have exercised due care if she or he leaves control to the autonomous system, as long as the autonomous system has accepted accountability. This principle can be embedded in the first design steps, either in a V-type or a more agile development process, e.g. using a spiral model with incremental refinement of the design and implementation

A fictitious case

Phase 1

Phase 2

Conclusions