Framework for Human-Automation Collaboration: Conclusions from Four Studies

Oxstrand, Johanna; Blanc, Katya Le; O’Hara, John M.; Joe, Jeffrey C.; Whaley, April M.; Medema, Heather

doi:10.2172/1123843

Cited by 3 publications

(3 citation statements)

References 85 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Attempts have also been made to create design guidelines which address human-system teamwork, collaboration, and task allocation (C. E. Billings, 1991;Donald A. Norman, 1993;Hilburn & Flynn, 2001;Glen Klein et al, 2004;O'Hara & Higgins, 2010;Oxstrand et al, 2013;D. Norman, 2017).…”

Section: Human-automation Teamwork Designmentioning

confidence: 99%

See 1 more Smart Citation

Human-automation teamwork : Current practices and future directions in air traffic control

Svensson

2020

Linköping Studies in Science and Technology. Dissertations

View full text Add to dashboard Cite

This dissertation explores the topic of human-automation teamwork in Air Traffic Control (ATC). ATC is a high stakes environment where complex automation is being introduced while the human operator has the legal responsibility. With increasing demands on productivity in various industries (as also in ATC), automation is introduced for efficiency, maintaining safety, and to keep the workload of the human operator within acceptable limits. However, previous research has shown that automation may cause negative effects on the human operator and performance, such as forcing the operator out of the control loop, which might lead to problems or confusion. Previous research suggests a need for strengthening human-automation collaboration where automation is seen as a team member to keep the operator in the loop. In order to achieve such teamwork, the design of the automation needs to be human-centred, i.e. that the automation is designed for the underlying need of the operator.The aim of this dissertation is to explore teamwork in ATC from several angles to understand how the air traffic controllers are working in current ATC environments and how automation could be designed to support humanautomation teamwork. The included studies rely on interviews, simulations, and questionnaires, all with operational air traffic controllers as participants.The results indicate that for both human-human teamwork and humanautomation teamwork, teamwork factors such as adaptability and mutual performance monitoring (knowing what the other team members are doing) are important for the work performance in current ATC environments, where mutual performance monitoring is especially important during stressful situations.When designing automation, lessons learned from human-human teamwork should be considered. The work within the scope of this dissertation identifies and concerns two human-automation teamwork aspects: boundary awareness and implicit communication. These are proposed to support the operator's knowledge about the automation and the communication flow between the iv operator and the automation. Boundary awareness is the operator's knowledge of the automation's abilities, its boundaries (what it can or cannot manage), and about consequences if it would go outside of these boundaries. Implicit communication is the unspoken or implied small cues that the operator and the automation can use to communicate with each other. It is proposed that implicit communication can be based on the work patterns of the operator. The knowledge gained through the work in this dissertation can be used as a foundation for further research and design of automation regarding operator knowledge about the automation boundaries and the communication within the team. v vii INCLUDED PUBLICATIONS

show abstract

Section: Human-automation Teamwork Designmentioning

confidence: 99%

“…Norman, 2017), and the differences and similarities between human-human and humanautomation teamwork (Nass, Fogg, & Moon, 1996;P. U. Lee, 2005;Oxstrand et al, 2013). Having human-automation teamwork in ATM might enhance performance while maintaining a high safety level.…”

Section: Towards Human-automation Teamworkmentioning

confidence: 99%

Human-automation teamwork : Current practices and future directions in air traffic control

Svensson

2020

Linköping Studies in Science and Technology. Dissertations

View full text Add to dashboard Cite

show abstract

“…The investigation of state-of-practice identified several important features of automation that influence the collaboration between automated systems and human operators, including; the level of automation (LOA), the cognitive function replaced by automation (e.g., detecting anomalies or taking control actions), adaptability, reliability, the process of automation, and automation modes. Oxstrand et al, 2013aOxstrand et al, , 2013bOxstrand et al, , and 2013c, provide more information related to the literature review and the development of the HAC model.…”

Section: Introductionmentioning

confidence: 99%

Enabling Situation Awareness Under High Levels of Automation: Results From An Experimental Study

Blanc

Spielman

Bower

et al. 2015

Self Cite

View full text Add to dashboard Cite

Automation poses a great opportunity to increase both system and human performance. However, it is important to pay attention to the design of the collaboration between the human operator and the automated systems. If not designed correctly, the overall performance may be severely reduced. Previous research indicates that an intermediate level of automation might be the solution that minimizes the tradeoffs between manual and automated actions. However, there are still many remaining unanswered questions related to impacts on situation awareness (SA) and workload during different levels of automation (LOAs). Some of the documented issues of automation include reduced SA, complacency, and over reliance on the system automation.The Advanced Reactor Technologies Human Automation Collaboration Research Project has conducted two experimental studies investigating the effect of different LOAs on system and human performance. The results from the first experimental study were published in Oxstrand and Le Blanc (2014) and Le Blanc and Oxstrand (2015) and are summarized in this report. The second experimental study was conducted by Idaho National Laboratory researchers. This report focuses on the second experimental study and details the design of the study and results.The second experimental study aimed to investigate the effects automation has on overall human-system performance. Researchers employed a simplified process control simulation using a nested experimental design to evaluate performance differences between four LOAs (manual, intermediate, adaptable, and automatic) with varying faults and conditional changes occurring during each scenario. The effects of automation were measured by participants' ability to detect and respond to failing automation and conditional changes using a variety of metrics assessing SA, system performance, and workload. The intent of this study is to determine the optimum conditions for human automation collaboration.The results of the present study indicate that more automation led to better system performance under normal operating conditions, and poorer SA and system performance under fault conditions. The results for the two middle LOAs are not as straightforward. Consistent with the hypothesis, system performance using adaptable automation exceeded system performance when using the intermediate LOA. However, contrary to the hypotheses, SA and fault performance were not superior using the adaptable LOA than using the intermediate LOA. The results of this study indicate that although adaptable automation fulfills half if its promise (i.e., it produces system performance closer to fully automatic than intermediate LOAs), it falls short of also enhancing SA and fault performance. This study demonstrates the fundamental tradeoff associated with high LOAs: automation enhances performance during normal conditions, but increases failure (sometimes catastrophically) of the humansystem under the condition of automation failure. In contrast, although manual performance is inferior to automati...

show abstract