Trust, control strategies and allocation of function in human-machine systems.

As automated controllers supplant human intervention in controlling complex systems, the operators' role often changes from that of an active controller to that of a supervisory controller. Acting as supervisors, operators can choose between automatic and manual control. Improperly allocating function between automatic and manual control can have negative consequences for the performance of a system. Previous research suggests that the decision to perform the job manually or automatically depends, in part, upon the trust the operators invest in the automatic controllers. This paper reports an experiment to characterize the changes in operators' trust during an interaction with a semi-automatic pasteurization plant, and investigates the relationship between changes in operators' control strategies and trust. A regression model identifies the causes of changes in trust, and a 'trust transfer function' is developed using time series analysis to describe the dynamics of trust. Based on a detailed analysis of operators' strategies in response to system faults we suggest a model for the choice between manual and automatic control, based on trust in automatic controllers and self-confidence in the ability to control the system manually.     

Factors affecting performance on a target monitoring task employing an automatic tracker

The experiments in this paper examined the extent to which performance on a task employing an automatic tracker was similar to performance on tasks employing other types of automation that have been studied more extensively. Automated target tracking is being used in many sensor and navigation systems to improve performance and help the operator cope with increased data loads. With many automated systems these goals are not met. In particular, the operator often misses errors made by the automated system and may report no decrease in workload. Several hypotheses have been offered for the operator's failure to monitor an automated system adequately. These include lack of experience with the manual task, a vigilance decrement, complacency, and inappropriate level of automation. The relevance of each of these hypotheses to failure to monitor an automatic tracker adequately was examined. Performance and perceived workload on a target tracking task employing an automatic tracker, in which participants had to detect and then update the position of several targets (e.g. ships) at regular intervals, were measured as a function of number of targets, training with the manual task, experience, and time on task. The results suggested that failure to detect errors made by the automated system was due largely to the lack of visibility of the automation errors relative to other errors. However, complacency could not be ruled out entirely. Unlike some other tasks, the availability of a reliable automatic tracker did lead to a substantial reduction in perceived workload.     

Task partitioning effects in semi-automated human–machine system performance

Twelve professional pilots performed a flight simulation consisting of three component sub-tasks: (i) tracking, (ii) monitoring and (iii) targeting, respectively. The targeting sub-task required (i) target identification, (ii) weapon selection and then (iii) weapon release. Pilots performed in a fully manual condition, a partial automation condition or a fully automated condition. Automated assistance was provided for the targeting sub-task only, while tracking and monitoring sub-tasks were always performed manually. During full automation, the computer located the target, identified it and released the appropriate weapon without any pilot input. During partial automation, the computer located and identified the target while the pilot retained final control over weapon release. Significantly higher levels of tracking error distinguished manual from both automated conditions and also between the two levels of automation. Monitoring response times were also sensitive to the degree of automation engaged, with the partial-automation condition exhibiting faster responses than full automation. Findings support a design principle in which pilots retain control over final weapons release directly on the basis of objective performance outcome. These collective results support the contention that effective and principled task-partitioning should represent a central strategy for the evolution of complex human–machine systems.

Practitioner Summary: Advantages of partitioning tasks between human and automated control are contingent upon the overall context of performance and the actual way the partitioning is accomplished. Simple algorithms, for example, automate on every feasible occasion, are poor design heuristics and may even prove actively harmful to overall response capacity. Transitioning humans from active controllers to passive monitors can be a problematic design choice, especially when that individual is socially deemed to retain overall responsibility for ultimate system effects in the real world.     

Extending the decision field theory to model operators' reliance on automation in supervisory control situations

Appropriate trust in and reliance on automation are critical for safe and efficient system operation. This paper fills an important research gap by describing a quantitative model of trust in automation. We extend decision field theory (DFT) to describe the multiple sequential decisions that characterize reliance on automation in supervisory control situations. Extended DFT (EDFT) represents an iterated decision process and the evolution of operator preference for automatic and manual control. The EDFT model predicts trust and reliance, and describes the dynamic interaction between operator and automation in a closed-loop fashion: the products of earlier decisions can transform the nature of later events and decisions. The simulation results show that the EDFT model captures several consistent empirical findings, such as the inertia of trust and the nonlinear characteristics of trust and reliance. The model also demonstrates the effects of different types of automation on ttrust and reliance. It is possible to expand the EDFT model for multioperator multiautomation situations.     

The effects of errors on system trust, self-confidence, and the allocation of control in route planning

The concept of trust is believed by some to compensate for feelings of uncertainty. Therefore, trust is considered to be crucial in people's decision to rely on a complex automated system to perform tasks for them. This experiment aimed to study the effects of errors on control allocation, and the mediating role of trust and self-confidence in the domain of route planning. Using a computer-based route planner, participants completed 10 route-planning trials in manual mode, and 10 in automatic mode, allowing participants to become equally experienced in operating both modes. During these so-called fixed trials, the numbers of errors in automatic as well as manual mode were systematically varied. Subsequently, participants completed six free trials, during which they were free to choose between modes. Our results showed that high automation error rates (AERs) decreased levels of system trust compared to low AERs. Conversely, high manual error rates (MERs) resulted in lower levels of self-confidence compared to low MERs, although to a lesser extent. Moreover, the difference between measures of trust and self-confidence proved to be highly predictive of the number of times automatic mode was selected during the six free trials. Additionally, results suggest a fundamental bias to trust one's own abilities over those of the system. Finally, evidence indicating a relationship between trust and self-confidence is discussed.     

The use of protocol analysis for determining ability requirements for personnel selection on a computer-based task

Abstract

The role of cognitively oriented tasks in the workplace continues to increase as automation of physical task components advances. Difficulties in automating the operator's cognitive processes have placed a renewed emphasis on the human component in advanced manufacturing systems. While traditional task analysis techniques have made significant contributions to improving productivity when important task elements are visually observable, their focus on manual task procedures make them less effective for cognitively oriented activities. This research has made a first attempt at integrating techniques from several disciplines to develop a cognitive task analysis methodology. The utility of this combined approach is examined for a new system being tested in the United States Postal Service. This task requires operators to encode, via a keyboard, addresses presented on a video display terminal. Results support the hypothesis that, for cognitively oriented tasks, a consensus based analysis technique (the Position Analysis Questionnaire) can be significantly improved by including data from task analysis provided the methodology is suitable for identifying non-physical task components.     

Predictive human operator model to be utilized as a controller using linear,neuro-fuzzy and fuzzy-ARX modeling techniques

Modeling human operator's behavior as a controller in a closed-loop control system recently finds applications in areas such as training of inexperienced operators by expert operator's model or developing warning systems for drivers by observing the driver model parameter variations. In this research, first, an experimental setup has been developed for collecting data from human operators as they controlled a nonlinear system. Appropriate reference signals and scenarios were designed according to the system identification and human operator modeling theory, to collect data from subjects. Different modeling schemes, namely ARX models as linear approach, and adaptive-network-based fuzzy inference system (ANFIS) as intelligent modeling approach have been evaluated. A hybrid modeling method, fuzzy-ARX (F-ARX) model, has been developed and its performance was found to be better in terms of predicting human operator's control actions as well as replacing the operator as a stand-alone controller. It has been concluded that F-ARX models can be a good alternative for modeling the human operator.     

Handing over the Keys: A Qualitative Study of the Experience of Automation in Driving

ABSTRACT

This article presents a qualitative driving simulator study designed to understand the experience of giving up control to automated processes in semiautonomous driving systems. The study employed an experience prototyping methodology, with 12 drivers (4 female) completing 2 sessions in a high-fidelity driving simulator. Condition A simulated a normally functioning car, while Condition B simulated a semiautonomous system that monitors driver behavior and takes evasive action when danger is detected. The simulator experience was used to ground wider discussion of automation and the experience of driving, which was explored through a semistructured interview. Results identify design challenges for autonomous driving systems; the loss of user agency and confidence, and handling the change between manual and automated control. Opportunities were identified; in augmenting rather than removing human abilities, and in providing new learning opportunities for drivers.     

A model for types and levels of human interaction with automation

We outline a model for types and levels of automation that provides a framework and an objective basis for deciding which system functions should be automated and to what extent. Appropriate selection is important because automation does not merely supplant but changes human activity and can impose new coordination demands on the human operator. We propose that automation can be applied to four broad classes of functions: 1) information acquisition; 2) information analysis; 3) decision and action selection; and 4) action implementation. Within each of these types, automation can be applied across a continuum of levels from low to high, i.e., from fully manual to fully automatic. A particular system can involve automation of all four types at different levels. The human performance consequences of particular types and levels of automation constitute primary evaluative criteria for automation design using our model. Secondary evaluative criteria include automation reliability and the costs of decision/action consequences, among others. Examples of recommended types and levels of automation are provided to illustrate the application of the model to automation design     

A graphic modeling and analysis tool for human fault diagnosis tasks

This study presents a graphic modeling and analysis tool for use in constructing an operator's mental model in fault diagnosis tasks. In most automatic and complicated process control systems, human fault diagnosis tasks have become increasingly complex and specialized. The system designer should consider the cognitive process of human operator to avert failure of implement action owing to a lack of compatibility between humans and aiding system interface. Here, an experiment is performed to investigate the nature of human fault diagnosis. A graphic modeling and analysis tool is then proposed to model the continuous process of human fault diagnosis. The approach proposed herein exploits both the line-chart and Petri nets to demonstrate the operator's thoughts and actions. Moreover, results in this study are integrated into an adaptive standard diagnosis model that can assess the operators' mental workload and accurately depict the interactions between human operator and aiding system.Relevance to industryAutomatic intelligent diagnosis systems can not provide satisfactory operating performance. Human diagnosticians are more effective than computer ones. Results in this study offer further insight into an operator behavior in graphic form and also how to design a better aiding system.     

How does people's trust in automated vehicles change after automation errors occur? An empirical study on dynamic trust in automated driving

When errors of automated vehicles (AVs) occur, drivers' trust can easily be destroyed, resulting in the reduction of the use of AVs. This study aims to examine how error of AVs declines driver's trust by impacting their subjective perceptions. A driving simulator experiment is conducted, in which 104 participants (male = 58; female = 46) experienced automated driving with automation errors and rated their trust. The results indicate that automation error will affect the driver's perceived predictability, perceived reliability, and perceived safety, which will lead to the decline of trust and abandonment of automated driving. With the occurrence of automation error of AVs, perceived safety plays a more critical role in drivers' trust. In addition, when automation errors occur in specific tasks with low risk, the trust of drivers will drop faster than that in high-risk tasks. This paper has explored the internal effects of the decline of driver's trust after automation errors of AVs, and further considers the influence of different external risks on these perception factors and trust. This study can help AVs manufacturers to formulate different degrees of trust repair strategies according to different driving tasks and accident severity.     

Human–Vehicle Cooperation in Automated Driving: A Multidisciplinary Review and Appraisal

ABSTRACT

To draw a comprehensive and cohesive understanding of human–vehicle cooperation in automated driving, a review is made on key studies in human–robot interaction and human factors. Throughout this article, insight is provided into how human drivers and vehicle systems interplay and influence each other. The limitations of technology-centered taxonomies of automation are discussed and the benefits of accounting for human agents are examined. The contributions of machine learning to automated driving and how critical models in human-system cooperation can inform the design of a more symbiotic relationship between driver and vehicle are investigated. Challenges in the human element to enable the safe introduction of road automation are also discussed. Particularly, the unintended consequences of vehicle automation on driver’s workload, situation awareness and trust are examined, and the social interactions between driver, vehicle, and other road users are investigated. This review will help professionals shape future directions for safer and more efficient and effective human–vehicle cooperation.     

Towards a dynamic balance between humans and automation: authority, ability, responsibility and control in shared and cooperative control situations

Progress enables the creation of more automated and intelligent machines with increasing abilities that open up new roles between humans and machines. Only with a proper design for the resulting cooperative human–machine systems, these advances will make our lives easier, safer and enjoyable rather than harder and miserable. Starting from examples of natural cooperative systems, the paper investigates four cornerstone concepts for the design of such systems: ability, authority, control and responsibility, as well as their relationship to each other and to concepts like levels of automation and autonomy. Consistency in the relations between these concepts is identified as an important quality for the system design. A simple graphical tool is introduced that can help to visualize the cornerstone concepts and their relations in a single diagram. Examples from the automotive domain, where a cooperative guidance and control of highly automated vehicles is under investigation, demonstrate the application of the concepts and the tool. Transitions in authority and control, e.g. initiated by changes in the ability of human or machine, are identified as key challenges. A sufficient consistency of the mental models of human and machines, not only in the system use but also in the design and evaluation, can be a key enabler for a successful dynamic balance between humans and machines.     

Detecting automation failures in a simulated supervisory control environment

Abstract

The goal of this research was to determine how individuals perform and allocate their visual attention when monitoring multiple automated displays that differ in automation reliability. Ninety-six participants completed a simulated supervisory control task where each automated display had a different level of reliability (namely 70%, 85% and 95%). In addition, participants completed a high and low workload condition. The performance data revealed that (1) participants’ failed to detect automation misses approximately 2.5 times more than automation false alarms, (2) participants’ had worse automation failure detection in the high workload condition and (3) participant automation failure detection remained mostly static across reliability. The eye tracking data revealed that participants spread their attention relatively equally across all three of the automated displays for the duration of the experiment. Together, these data support a system-wide trust approach as the default position of an individual monitoring multiple automated displays.

Practitioner Summary: Given the rapid growth of automation throughout the workforce, there is an immediate need to better understand how humans monitor multiple automated displays concurrently. The data in this experiment support a system-wide trust approach as the default position of an individual monitoring multiple automated displays.

Abbreviations: DoD: Department of Defense; UA: unmanned aircraft; SCOUT: Supervisory Control Operations User Testbed; UAV: unmanned aerial vehicle; AOI: areas of interest     

The role of trust in automation reliance

A recent and dramatic increase in the use of automation has not yielded comparable improvements in performance. Researchers have found human operators often underutilize (disuse) and overly rely on (misuse) automated aids (Parasuraman and Riley, 1997). Three studies were performed with Cameron University students to explore the relationship among automation reliability, trust, and reliance. With the assistance of an automated decision aid, participants viewed slides of Fort Sill terrain and indicated the presence or absence of a camouflaged soldier. Results from the three studies indicate that trust is an important factor in understanding automation reliance decisions. Participants initially considered the automated decision aid trustworthy and reliable. After observing the automated aid make errors, participants distrusted even reliable aids, unless an explanation was provided regarding why the aid might err. Knowing why the aid might err increased trust in the decision aid and increased automation reliance, even when the trust was unwarranted. Our studies suggest a need for future research focused on understanding automation use, examining individual differences in automation reliance, and developing valid and reliable self-report measures of trust in automation.     

基于机器视觉的嫁接夹抓取点定位与朝向检测

针对现代农业自动化嫁接机的嫁接过程中,嫁接夹的分拣整理仍然需要使用刚性振动盘或人工参与,整体自动化程度有待提高的问题,提出基于机器视觉的嫁接夹轮廓链特征点定位方法与朝向检测;设计了一种自动化嫁接专用的嫁接夹,采用背光源打光方式,对每个阴影轮廓计算Hu矩,根据Hu矩匹配值进行分类以获得正确的嫁接夹个体轮廓;对每个嫁接夹个体轮廓采用轮廓链角分析方法,获得嫁接夹抓取点的像素位置;根据抓取点位置信息计算朝向角度;实验结果表明,嫁接夹抓取点平均定位误差为2.16个像素,嫁接夹朝向角度检测误差为0.40°;该方法可以准确地对嫁接夹进行视觉检测,对嫁接机的嫁接过程中,嫁接夹的自动分拣方面,具有机器视觉方面的价值。     

Comparison of a brain-based adaptive system and a manual adaptable system for invoking automation

OBJECTIVE: Two experiments are presented examining adaptive and adaptable methods for invoking automation. BACKGROUND: Empirical investigations of adaptive automation have focused on methods used to invoke automation or on automation-related performance implications. However, no research has addressed whether performance benefits associated with brain-based systems exceed those in which users have control over task allocations. METHOD: Participants performed monitoring and resource management tasks as well as a tracking task that shifted between automatic and manual modes. In the first experiment, participants worked with an adaptive system that used their electroencephalographic signals to switch the tracking task between automatic and manual modes. Participants were also divided between high- and low-reliability conditions for the system-monitoring task as well as high- and low-complacency potential. For the second experiment, participants operated an adaptable system that gave them manual control over task allocations. RESULTS: Results indicated increased situation awareness (SA) of gauge instrument settings for individuals high in complacency potential using the adaptive system. In addition, participants who had control over automation performed more poorly on the resource management task and reported higher levels of workload. A comparison between systems also revealed enhanced SA of gauge instrument settings and decreased workload in the adaptive condition. CONCLUSION: The present results suggest that brain-based adaptive automation systems may enhance perceptual level SA while reducing mental workload relative to systems requiring user-initiated control. APPLICATION: Potential applications include automated systems for which operator monitoring performance and high-workload conditions are of concern.     

An Architecture for Human‐Guided Autonomy: Team TROOPER at the DARPA Robotics Challenge Finals

Recent robotics efforts have automated simple, repetitive tasks to increase execution speed and lessen an operator's cognitive load, allowing them to focus on higher‐level objectives. However, an autonomous system will eventually encounter something unexpected, and if this exceeds the tolerance of automated solutions, there must be a way to fall back to teleoperation. Our solution is a largely autonomous system with the ability to determine when it is necessary to ask a human operator for guidance. We call this approach human‐guided autonomy. Our design emphasizes human‐on‐the‐loop control where an operator expresses a desired high‐level goal for which the reasoning component assembles an appropriate chain of subtasks. We introduce our work in the context of the DARPA Robotics Challenge (DRC) Finals. We describe the software architecture Team TROOPER developed and used to control an Atlas humanoid robot. We employ perception, planning, and control automation for execution of subtasks. If subtasks fail, or if changing environmental conditions invalidate the planned subtasks, the system automatically generates a new task chain. The operator is able to intervene at any stage of execution, to provide input and adjustment to any control layer, enabling operator involvement to increase as confidence in automation decreases. We present our performance at the DRC Finals and a discussion about lessons learned.     

Investigating the effects of experience on human performance in an object-tracking task: a case study of manual rendezvous and docking

Human involvement with the manual control of object-tracking tasks, such as manual rendezvous and docking, can enhance the flexibility of the task. The operator is required to make decisions as precisely as possible when a chase vehicle is approaching the target vehicle. As the level of mental workload is intensively increased, the operator's perception and decision-making with tracking error-correcting purposes should be properly identified in multiple degrees of freedom (DOFs). The two experimental studies in this paper were conducted for object-based selective attention in six DOFs and object-pointing manipulation under the zero resistance circumstance, which were correlated with human performance when a chaser approached a target spacecraft; the effects of experience level on the two important factors are presented herein. The experimental results demonstrated that some performance details between the novice and experienced groups were different in terms of the attentional priority of multiple DOFs and lateral/vertical movement for deviation error correction, which could provide a certain reference for a cognitive–behavioural model and assist the operator in obtaining proper or even optimal decision-making or operating instructions to improve their adaptation to selective attention and object-pointing manipulation, for preventing accidental manipulation and performance degradation.     

Performance evaluation of automated brain tumor detection systems with expert delineations and interobserver variability analysis in diseased patients on magnetic resonance imaging

ABSTRACT

Intervention by human expert has turned out to be essential for computerized analysis systems desiring to be approved by medical regulatory bodies. Further, to validate the performance of automated diagnosis systems, interobserver variability analysis is critically important. The purpose of this article is twofold: (i) firstly to perform interobserver variability analysis of two experienced Radiologists interpreting lesion boundary in brain magnetic resonance images; (ii) secondly, to evaluate the performance of recently proposed automated lesion segmentation model with that of the two experienced Radiologists who performed manual delineations of lesion boundary. Experiments were conducted on the database consisting of 80 real-time brain images with glioma tumor acquired using magnetic resonance imaging (MRI). Extensive statistical analysis such as the two tailed T-test, analysis of variance (ANOVA) test, Mann-Whitney U test, regression and correlation tests, etc. are performed to compare the lesions detected manually by experts and that by the automated method. Furthermore, three quantitative measures namely, dice similarity index, Jaccard coefficient, and Hausdorff distance are used to evaluate the automated lesion detection method. The experimental results show that the lesion boundaries detected by the automated method are very close to the manual delineations provided by the expert Radiologists. It is concluded that the automated systems for brain lesion detection can be utilized as a part of routine clinical practice to help the medical professionals in determining the exact location and area of lesions in brain MRI images.