Human-in-the-Loop and Automation Transparency in Remote Monitoring¶

Status: established
Last updated: 2026-05-31
Sources: Poratheafhe2022.Pdf, Porathe2021_Chapter_Autonomousshipsaresearchstrate.Pdf
Tags: [human-in-the-loop, out-of-the-loop, automation-transparency, qgild, situational-awareness, remote-monitoring, mass, roc]

Summary¶

Remote monitoring of autonomous ships creates a specific failure mode: an operator who watches well-functioning automation for long periods drifts "out of the loop" and cannot rapidly regain the awareness needed to intervene when a ship suddenly requires assistance (Porathe, 2022). Porathe frames this as the central design problem for Remote Operation Centres and proposes building automation transparency into the human-machine interface, including a Quickly-Getting-Into-the-Loop Display (QGILD) intended to compress the time an operator needs to re-enter the loop. The problem connects directly to situation awareness and to the broader research agenda for MASS Human Factors.

Body¶

Context¶

This article draws on two works by Porathe (2021, 2022) addressing how a shore operator stays able to intervene in autonomous ships. The research-strategy chapter (2021) sets out an eight-task Human Factors agenda for MASS, and the QGILD work-in-progress paper (2022) develops the out-of-the-loop problem and a proposed display response. Both treat remote monitoring as the setting and the operator's readiness to take over as the central design problem. The article is the human-readiness strand of this knowledge base: it sharpens the monitoring half of the Remote Operation Centre defined in Remote Operation Centres Mass, shares its workload and vigilance concerns with Multi Ship Remote Operations Workload Sa, and connects transparency to the calibrated-trust argument in Trust In Human Autonomy Teaming.

Key Points¶

The core tension is that automation working well most of the time degrades human readiness to take over when it fails. Porathe (2022) describes the out-of-the-loop syndrome, drawing on Endsley and Jones: an operator monitoring well-functioning automation begins thinking about other things and slowly slides out of the loop. While the automation is working, being out of the loop carries little immediate cost; the danger surfaces when a monitored ship, after a long period of correct automatic operation, suddenly needs remote assistance and the operator is slow to detect the problem and must rebuild awareness fast (PDF pp. 3–4, orig. pp. 692–693). This makes the Human-In-The-Loop versus Human-Out-Of-The-Loop distinction a defining variable for future MASS development.

The stated goal is to build automation transparency into the ROC interface so operators can understand what the automation is doing and why. The QGILD paper (2022) frames this as its research question — when an operator is summoned to the console by an alarm without knowing what is going on, how can the interface bring them back in the loop as quickly as possible — and presents early concepts for the display (PDF p. 5, orig. p. 694). The same idea appears in the research-strategy chapter (2021), which lists a Quickly-Getting-Into-the-Loop Display (Research Task 4) and "The Glass Box: Automation Transparency" (Research Task 6) as distinct research tasks (PDF pp. 6–7, orig. pp. 483–484), alongside an Early Warning Look-Ahead function that would dynamically update operators well in advance to avoid surprises (PDF p. 6, orig. p. 483). The intent is to convert passive monitoring into a state from which active control can be resumed with minimal delay.

These proposals respond to a documented weakness in human supervisory control: Porathe (2021) notes that research shows humans are very bad at monitoring well-functioning automation, so active monitoring can lapse into passive monitoring, which is why keeping the operator alert and in the loop — rather than merely present — is the central problem of ROC staffing, and why he proposes the Operator Readiness Level concept (PDF p. 6, orig. p. 483). The MUNIN-era practice of having operators spend about ten minutes "virtually onboard" each ship once an hour was an early procedural attempt at the same goal, forcing periodic re-engagement to prevent loop-out (see Remote Operation Centres Mass).

Automation transparency sits within the eight-task Human Factors agenda, linking it to handover and to interaction with conventional ships. Research Task 7 addresses human intervention and handover (PDF p. 7, orig. p. 484), and Research Task 3 addresses situation awareness in remote monitoring, with information access identified as a research priority (PDF p. 6, orig. p. 483). The full eight-task list is set out in the chapter's conclusion (PDF p. 8, orig. p. 485). Transparency, look-ahead warning, and rapid loop re-entry are thus components of one problem.

Conclusion¶

Porathe (2021, 2022) frames the out-of-the-loop problem as the defining design challenge for Remote Operation Centres: the better the automation, the less ready the human is to resume control. His proposed response is to make the automation transparent and to compress loop re-entry through the QGILD and an early-warning look-ahead, so the operator can take over meaningful control the moment automation reaches its limits. The proposals remain conceptual — QGILD is presented as work-in-progress with no reported empirical evaluation of loop-re-entry time.

References¶

Porathe, T. (2021) 'Autonomous Ships: A Research Strategy for Human Factors Research in Autonomous Shipping', in Stanton, N. (ed.) Advances in Human Aspects of Transportation (AHFE 2021), LNNS 270. Cham: Springer, pp. 479-486. doi: 10.1007/978-3-030-80012-3_55. porathe2021strategy

Porathe, T. (2022) 'Remote Monitoring of Autonomous Ships: A Quickly Getting into the Loop Display (QGILD)', Advances in Transportation, 60, pp. 691-697. doi: 10.54941/ahfe1002506. porathe2022qgild

Open Questions¶

QGILD is presented as work-in-progress; no empirical evaluation of loop-re-entry time is reported in the source.
How does automation transparency interact with calibrated trust? See Trust In Human Autonomy Teaming.