Alexander J. Stimpson

Analysis of human spatial perception during lunar landing

Crewed lunar landings require astronauts to interact with automated systems to identify a location that is level and free of hazards and to guide the vehicle to the lunar surface through a controlled descent. However, vestibular limitations resulting from exposure to lunar gravity after short-term adaptation to weightlessness, combined with acceleration profiles unique to lunar landing trajectories may result in astronaut spatial disorientation. A quantitative mathematical model of human spatial orientation previously developed was adopted to analyze disorientation concerns during lunar landing conditions that cannot be reproduced experimentally. Vehicle acceleration and rotation rate profiles of lunar landing descent trajectories were compiled and entered as inputs to the orientation model to predict astronaut perceived orientations. Both fully automated trajectories and trajectories with pilot interaction were studied. The latter included both simulated landing point redesignation and direct manual control. The lunar descent trajectories contain acceleration and rotation rate profiles producing attitude perceptions that differ substantially from the actual vehicle state. In particular, a somatogravic illusion is predicted that causes the perceived orientation to be nearly upright compared to the actual vehicle state which is pitched back. Furthermore, astronaut head location within the vehicle is considered for different vehicle designs to determine the effect on perceived orientation. The effect was found to be small, but measureable (0.3-4.1 degrees), and larger for the new Altair vehicle design compared to the Apollo Lunar Module.

Assessing Intervention Timing in Computer-Based Education Using Machine Learning Algorithms

The use of computer-based and online education systems has made new data available that can describe the temporal and process-level progression of learning. To date, machine learning research has not considered the impacts of these properties on the machine learning prediction task in educational settings. Machine learning algorithms may have applications in supporting targeted intervention approaches. The goals of this paper are to: 1) determine the impact of process-level information on machine learning prediction results and 2) establish the effect of type of machine learning algorithm used on prediction results. Data were collected from a university level course in human factors engineering (n=35), which included both traditional classroom assessment and computer-based assessment methods. A set of common regression and classification algorithms were applied to the data to predict final course score. The overall prediction accuracy as well as the chronological progression of prediction accuracy was analyzed for each algorithm. Simple machine learning algorithms (linear regression, logistic regression) had comparable performance with more complex methods (support vector machines, artificial neural networks). Process-level information was not useful in post-hoc predictions, but contributed significantly to allowing for accurate predictions to be made earlier in the course. Process level information provides useful prediction features for development of targeted intervention techniques, as it allows more accurate predictions to be made earlier in the course. For small course data sets, the prediction accuracy and simplicity of linear regression and logistic regression make these methods preferable to more complex algorithms.

Functional Requirements for Onboard Intelligent Automation in Single Pilot Operations

There is growing interest in the concept of Single Pilot Operations (SPO) within commercial flight operations due to the potential economic benefits. Prior research has focused on architectures and safety concerns related to SPO, but has not examined what functionalities automation would need to fulfill in the replacement of a co-pilot. Through guided interviews conducted with experienced commercial airline pilots, this effort demonstrates what functionalities would need to be replicated by an on-board intelligent system. These interviews revealed both desired and potentially deficient qualities of co-pilots, providing some guidance into how automated systems could be designed to best replace current human co-pilots. The interviews also provided perspectives about the issues of pilot selection and training, and the social implications of the use of an automated system in the cockpit rather than a human co-pilot. Given the results from the interviews, we developed a list of a dozen functionalities and capabilities that an onboard intelligent system should be able to replicate in order for a single human pilot to be able to manage the workload in piloting an aircraft in transport missions.

Exploring Concepts of Operations for On-Demand Passenger Air Transportation

In recent years, a surge of interest in “flying cars” for city commutes has led to rapid development of new technologies to help make them and similar on-demand mobility platforms a reality. To this end, this paper provides analyses of the stakeholders involved, their proposed operational concepts, and the hazards and regulations that must be addressed. Three system architectures emerged from the analyses, ranging from conventional air taxi to revolutionary fully autonomous aircraft operations, each with vehicle safety functions allocated differently between humans and machines. Advancements for enabling technologies such as distributed electric propulsion and artificial intelligence have had major investments and initial experimental success, but may be some years away from being deployed for on-demand passenger air transportation at scale.

Assessing Pilot Workload in Single-Pilot Operations with Advanced Autonomy

The proposed transition to single-pilot operations (SPO) in commercial and military aircraft has motivated the development of advanced autonomy systems. However, a detailed analysis of the impact of advanced autonomy on pilot workload through various phases of flight and contingency scenarios has not been conducted. To this end, this paper presents the development of the Pilot-Autonomy Workload Simulation (PAWS), a discrete event simulation model that allows the investigation of pilot workload under a variety of advanced autonomy capabilities and scenarios. Initial utilization results from PAWS of nominal and off-nominal point-to-point missions demonstrate that the workload for a single pilot assisted by advanced autonomy varies considerably over different phases of flight and various contingencies. These results suggest that advanced autonomy to offset pilot workload is not needed for low-workload phases, but could be critical during periods of high workload.

Human Spatial Orientation Perception During Simulated Lunar Landing Motions

Safe and precise piloted lunar landings require control inputs that depend on an accurate perception of vehicle orientation and motion. However, the unique environment and motions experienced during a lunar landing trajectory may lead to misperceptions in vehicle state. Eight subjects participated in a human subject experiment in the NASA Ames vertical motion simulator, where self-reports of perceptions of vehicle tilt angle and horizontal velocity were made during lunar-landing-like motions. Three cases of sensory cues were studied: Subjects were blindfolded and given no visual cues; subjects were provided a simulated dynamic view of the lunar terrain out a forward-looking window; and subjects were provided dynamic instrument displays showing current vehicle states. Subjects’ perception indications differed substantially from the motions being simulated in the blindfolded and out-the-window conditions, but were better matched when viewing instrument displays. Subject perceptions were also compared with p...

Functional Requirements for Remotely Managing Fleets of On-Demand Passenger Aircraft

The concept of On-Demand Mobility (ODM) in aviation has gained popularity in recent years, with several manufacturers proposing vehicles for high-speed intra-city air taxis. However, less attention has been placed on how these fleets would be operationally controlled and managed. Through the development of concepts of operations for remote management of vehicles with differing levels of autonomy, this paper presents preliminary requirements for ODM air operations control centers. The centers would interface with air traffic control and be responsible for ensuring safe, efficient, and effective operations of fleets within subareas of the National Airspace System. Our effort identified key functional requirements related to vehicle safety, customer experience, and airspace integration for these futuristic concepts. Further, this work introduces a novel Remote Operations Center (ROC) concept with highly integrated human-machine systems for efficient operations with limited staffing. The ROC would support the transition from providing dispatcher-like support to supervisory control of autonomous ODM systems, including managing emergencies, which will be crucial for operational success as vehicle autonomy evolves.

A Model-Based Measure to Assess Operator Adherence to Procedures

Procedures play an important role in domains where humans interact with critical, complex systems. In such environments, the operator’s ability to correctly follow a given set of procedures can directly impact system safety. A quantitative measure of procedural adherence during training for complex system operation would be useful to assess trainee performance and evaluate a training program. This paper presents a novel model-based objective metric for quantifying procedural adherence in training. This metric is sensitive to both the number and nature of procedural deviations, and can be used with cluster analysis to classify trainee performance based on adherence. The metric was tested on an experimental data set gathered from volunteers using aircraft maintenance computer-based training (CBT). The properties of the metric are discussed, along with future possibilities.

Effects of an achievability display during simulated lunar landings

Landing on the moon requires the selection and identification of a location that is level and free of hazards, along with a stable, controlled descent to the lunar surface through the use of automated systems and manual control. The pilot workload associated with both selecting a suitable landing point (LP) and monitoring vehicle state is a concern for future lunar or planetary landings. A novel achievability contour display element was developed to show the current achievable limit of the vehicle. A subject experiment was conducted in a lunar landing simulation environment to test the effects of the achievability contour on pilot performance, situation awareness, and workload in simulated approach and terminal descent scenarios as compared to an Apollo-style auditory display. Two control modes were used: supervisory control and roll, pitch, and yaw rate-control/attitude-hold (RCAH) manual control. The experiment also investigated differences in display effect with and without a required redesignation. Results of the subject experiment (n = 10) indicate that the achievability contour display showed significant improvement in subjective situation awareness and workload ratings. The results also indicate changes in the landing point selection process with the use of the achievability contours. There was no measurable difference in flight and landing performance measures between the two display conditions. The results of the experiment have shown that providing the achievability contour display has beneficial effects on pilot situation awareness and workload during the final approach and terminal descent maneuvers. Additional research is needed to determine the optimal implementation and pilot interaction methods in the use of this display.