Revising human-systems engineering principles for embedded AI applications.

The recent shift from predominantly hardware-based systems in complex settings to systems that heavily leverage non-deterministic artificial intelligence (AI) reasoning means that typical systems engineering processes must also adapt, especially when humans are direct or indirect users. Systems with embedded AI rely on probabilistic reasoning, which can fail in unexpected ways, and any overestimation of AI capabilities can result in systems with latent functionality gaps. This is especially true when humans oversee such systems, and such oversight has the potential to be deadly, but there is little-to-no consensus on how such system should be tested to ensure they can gracefully fail. To this end, this work outlines a roadmap for emerging research areas for complex human-centric systems with embedded AI. Fourteen new functional and tasks requirement considerations are proposed that highlight the interconnectedness between uncertainty and AI, as well as the role humans might need to play in the supervision and secure operation of such systems. In addition, 11 new and modified non-functional requirements, i.e., "ilities," are provided and two new "ilities," auditability and passive vulnerability, are also introduced. Ten problem areas with AI test, evaluation, verification and validation are noted, along with the need to determine reasonable risk estimates and acceptable thresholds for system performance. Lastly, multidisciplinary teams are needed for the design of effective and safe systems with embedded AI, and a new AI maintenance workforce should be developed for quality assurance of both underlying data and models.

The need for command and control instant message adaptive interfaces: lessons learned from Tactical Tomahawk human-in-the-loop simulations.

In the recent development of a human-in-the-loop simulation test bed designed to examine human performance issues for supervisory control of the Navy's new Tactical Tomahawk missile, measurements of operator situation awareness (SA) and workload through secondary tasking were taken through an embedded instant messaging program. Instant message interfaces (otherwise known as "chat"), already a means of communication between Navy ships, allow researchers to query users in real-time in a natural, ecologic setting, and thus provide more realistic and unobtrusive measurements. However, in the course of this testing, results revealed that some subjects fixated on the real-time instant messaging secondary task instead of the primary task of missile control, leading to the overall degradation of mission performance as well as a loss of SA. While this research effort was the first to quantify command and control performance degradation as a result of instant messaging, the military has recognized that in its network centric warfare quest, instant messaging is a critical informal communication tool, but has associated problems. Recently, a military spokesman said that managing chat in current military operations was sometimes a "nightmare," because military personnel have difficulty in handling large amounts of information through chat, and then synthesizing knowledge from this information. This research highlights the need for further investigation of the role of instant messaging interfaces both on task performance and situation awareness, and how the associated problems could be ameliorated through adaptive display design.

Supervisory-level interruption recovery in time-critical control tasks.

This paper investigates the effectiveness of providing interruption recovery assistance in the form of an interactive visual timeline of historical events on a peripheral display in support of team supervision in time-critical settings. As interruptions can have detrimental effects on task performance, particularly in time-critical work environments, there is growing interest in the design of tools to assist people in resuming their pre-interruption activity. A user study was conducted to evaluate the use of an interactive event timeline that provides assistance to human supervisors in time-critical settings. The study was conducted in an experimental platform that emulated a team of operators and a mission commander performing a time-critical unmanned aerial vehicle (UAV) mission. The study results showed that providing interruption assistance enabled people to recover from interruptions faster and more accurately. These results have implications for interface design that could be adopted in similar time-critical environments such as air-traffic control, process control, and first responders.

Combining scanning tunneling microscopy and synchrotron radiation for high-resolution imaging and spectroscopy with chemical, electronic, and magnetic contrast.

The combination of high-brilliance synchrotron radiation with scanning tunneling microscopy opens the path to high-resolution imaging with chemical, electronic, and magnetic contrast. Here, the design and experimental results of an in-situ synchrotron enhanced x-ray scanning tunneling microscope (SXSTM) system are presented. The system is designed to allow monochromatic synchrotron radiation to enter the chamber, illuminating the sample with x-ray radiation, while an insulator-coated tip (metallic tip apex open for tunneling, electron collection) is scanned over the surface. A unique feature of the SXSTM is the STM mount assembly, designed with a two free-flex pivot, providing an angular degree of freedom for the alignment of the tip and sample with respect to the incoming x-ray beam. The system designed successfully demonstrates the ability to resolve atomic-scale corrugations. In addition, experiments with synchrotron x-ray radiation validate the SXSTM system as an accurate analysis technique for the study of local magnetic and chemical properties on sample surfaces. The SXSTM system's capabilities have the potential to broaden and deepen the general understanding of surface phenomena by adding elemental contrast to the high-resolution of STM.

The role of human-automation consensus in multiple unmanned vehicle scheduling.

OBJECTIVE: This study examined the impact of increasing automation replanning rates on operator performance and workload when supervising a decentralized network of heterogeneous unmanned vehicles. BACKGROUND: Futuristic unmanned vehicles systems will invert the operator-to-vehicle ratio so that one operator can control multiple dissimilar vehicles connected through a decentralized network. Significant human-automation collaboration will be needed because of automation brittleness, but such collaboration could cause high workload. METHOD: Three increasing levels of replanning were tested on an existing multiple unmanned vehicle simulation environment that leverages decentralized algorithms for vehicle routing and task allocation in conjunction with human supervision. RESULTS: Rapid replanning can cause high operator workload, ultimately resulting in poorer overall system performance. Poor performance was associated with a lack of operator consensus for when to accept the automation's suggested prompts for new plan consideration as well as negative attitudes toward unmanned aerial vehicles in general. Participants with video game experience tended to collaborate more with the automation, which resulted in better performance. CONCLUSION: In decentralized unmanned vehicle networks, operators who ignore the automation's requests for new plan consideration and impose rapid replans both increase their own workload and reduce the ability of the vehicle network to operate at its maximum capacity. APPLICATION: These findings have implications for personnel selection and training for futuristic systems involving human collaboration with decentralized algorithms embedded in networks of autonomous systems.

Auditory decision aiding in supervisory control of multiple unmanned aerial vehicles.

OBJECTIVE: This article is an investigation of the effectiveness of sonifications, which are continuous auditory alerts mapped to the state of a monitored task, in supporting unmanned aerial vehicle (UAV) supervisory control. BACKGROUND: UAV supervisory control requires monitoring a UAV across multiple tasks (e.g., course maintenance) via a predominantly visual display, which currently is supported with discrete auditory alerts. Sonification has been shown to enhance monitoring performance in domains such as anesthesiology by allowing an operator to immediately determine an entity's (e.g., patient) current and projected states, and is a promising alternative to discrete alerts in UAV control. However, minimal research compares sonification to discrete alerts, and no research assesses the effectiveness of sonification for monitoring multiple entities (e.g., multiple UAVs). METHOD: The authors conducted an experiment with 39 military personnel, using a simulated setup. Participants controlled single and multiple UAVs and received sonifications or discrete alerts based on UAV course deviations and late target arrivals. RESULTS: Regardless of the number of UAVs supervised, the course deviation sonification resulted in reactions to course deviations that were 1.9 s faster, a 19% enhancement, compared with discrete alerts. However, course deviation sonifications interfered with the effectiveness of discrete late arrival alerts in general and with operator responses to late arrivals when supervising multiple vehicles. CONCLUSIONS: Sonifications can outperform discrete alerts when designed to aid operators to predict future states of monitored tasks. However, sonifications may mask other auditory alerts and interfere with other monitoring tasks that require divided attention. APPLICATIONS: This research has implications for supervisory control display design.

Developing operator capacity estimates for supervisory control of autonomous vehicles.

OBJECTIVE: This study examined operators' capacity to successfully reallocate highly autonomous in-flight missiles to time-sensitive targets while performing secondary tasks of varying complexity. BACKGROUND: Regardless of the level of autonomy for unmanned systems, humans will be necessarily involved in the mission planning, higher level operation, and contingency interventions, otherwise known as human supervisory control. As a result, more research is needed that addresses the impact of dynamic decision support systems that support rapid planning and replanning in time-pressured scenarios, particularly on operator workload. METHOD: A dual screen simulation that allows a single operator the ability to monitor and control 8, 12, or 16 missiles through high level replanning was tested on 42 U.S. Navy personnel. RESULTS: The most significant finding was that when attempting to control 16 missiles, participants' performance on three separate objective performance metrics and their situation awareness were significantly degraded. CONCLUSION: These results mirror studies of air traffic control that demonstrate a similar decline in performance for controllers managing 17 aircraft as compared with those managing only 10 to 11 aircraft. Moreover, the results suggest that a 70% utilization (percentage busy time) score is a valid threshold for predicting significant performance decay and could be a generalizable metric that can aid in manning predictions. APPLICATION: This research is relevant to human supervisory control of networked military and commercial unmanned vehicles in the air, on the ground, and on and under the water.

Effects of single versus multiple warnings on driver performance.

OBJECTIVE: To explore how a single master alarm system affects drivers' responses when compared with multiple, distinct warnings. BACKGROUND: Advanced driver warning systems are intended to improve safety, yet inappropriate integration may increase the complexity of driving, especially in high workload situations. This study investigated the effects of auditory alarm scheme, reliability, and collision event type on driver performance. METHOD: Using a 2 x 2 x 4 mixed factorial design, we investigated the impact of two alarm schemes (master vs. individual) and two levels of alarm reliability (high and low) on distracted drivers' performance across four collision event types (frontal collision warnings, left and right lane departure warnings, and warnings for a fast-approaching following vehicle). RESULTS: Participants' reaction times and accuracy rates were significantly affected by the type of collision event and alarm reliability. The use of individual alarms, rather than a single master alarm, did not significantly affect driving performance in terms of reaction time or response accuracy. CONCLUSION: Even though a master alarm is a relatively uninformative warning, it produced statistically no different reaction times or accuracy results when compared with information-rich auditory icons, some of which were spatially located. In addition, unreliable alarms negatively impacted driver performance, regardless of event type or alarm scheme. APPLICATION: These results have important implications for the development and implementation of multiple driver warning systems.

Large community outbreak of cryptosporidiosis due to contamination of a filtered public water supply.

Between January 12 and February 7, 1987, an outbreak of gastroenteritis affected an estimated 13,000 people in a county of 64,900 residents in western Georgia. Cryptosporidium oocysts were identified in the stools of 58 of 147 patients with gastroenteritis (39 percent) tested during the outbreak. Studies for bacterial, viral, and other parasitic pathogens failed to implicate any other agent. In a random telephone survey, 299 of 489 household members exposed to the public water supply (61 percent) reported gastrointestinal illness, as compared with 64 of 322 (20 percent) who were not exposed (relative risk, 3.1; 95 percent confidence interval, 2.4 to 3.9). The prevalence of IgG to cryptosporidium was significantly higher among exposed respondents to the survey who had become ill than among nonresident controls. Cryptosporidium oocysts were identified in samples of treated public water with use of a monoclonal-antibody test. Although the sand-filtered and chlorinated water system met all regulatory-agency quality standards, sub-optimal flocculation and filtration probably allowed the parasite to pass into the drinking-water supply. Low-level cryptosporidium infection in cattle in the watershed and a sewage overflow were considered as possible contributors to the contamination of the surface-water supply. We conclude that current standards for the treatment of public water supplies may not prevent the contamination of drinking water by cryptosporidium, with consequent outbreaks of cryptosporidiosis.