Organizational Unit:

Daniel Guggenheim School of Aerospace Engineering

Permanent Link

https://hdl.handle.net/1853/70742

Parent Organization

Organizational Unit

College of Engineering

Includes Organization(s)

Organizational Unit

Aerospace Design Group

Organizational Unit

Aerospace Systems Design Laboratory (ASDL)

Organizational Unit

Air Transportation Laboratory

Organizational Unit

Cognitive Engineering Center

Organizational Unit

Space Systems Design Laboratory (SSDL)

Show 1 more

ArchiveSpace Name Record

https://finding-aids.library.gatech.edu/agents/corporate_entities/106

Full item page

Publication Search Results

Now showing 1 - 9 of 9

Understanding and Supporting Decision Making in Denied and Degraded Environments

(Georgia Institute of Technology, 2023-07-25) Sealy, William I.N.

Decision making is not guaranteed to occur in well-structured environments with perfect information. Tasks in the research most often focus on decisions made with complete information in an unlimited time-frame, and in cases where information is missing or uncertain, the current research stops short of addressing the effect of the distribution of the missing information in the environment. This dissertation seeks specifically to understand how these distributions of information affect decision makers under time pressure, and how best to support decision making in imperfect environments across a range of decision strategies. The contributions of the work are three fold. First, results showed that three studied factors of information distributions (namely Total Information, Complete Attribute Pairs, and Information Imbalance) were significant predictors of decision accuracy in six separate human subject studies featuring varying information complexity and decision strategy biases. Second, this dissertation has highlighted key differences in expert and novice behavior through the lens of information estimation and predecisional information search which further explained individual differences in performance under uncertainty and provided novel design considerations for decision support systems (DSS) in these environments. Finally, the application of both information modification and option prediction DSS showed significant increases in accuracy and reduction in response times across performance groups in both heuristic and analytically-biased environments.
Augmented reality cueing methodologies for rotorcraft shipboard landings

(Georgia Institute of Technology, 2022-06-22) Walters, Robert

Augmented Reality Cueing Methodologies For Rotorcraft Shipboard Landings Robert Walters 159 Pages The helicopter-ship interface is one of the most challenging flight regimes in which pilots operate. Several factors make this flight regime complicated, such as the relative motion between the aircraft, the ship, and the sea, and also the air wake turbulence and the confined nature of the landing zone. Degraded visuals conditions such as sea spray, adverse weather, and poor lighting conditions compound the other difficulties. The high pilot workload from these factors can lead to a loss of situational awareness which can result in catastrophic aircraft accidents. Currently fielded cueing systems are not up to this challenge. To reduce pilot workload and improve situational awareness and performance, better pilot cueing is required. This dissertation investigated the extent to which augmented reality cueing utilizing modern rendering techniques reduces pilot workload and improves situational awareness and performance. This was done by supporting a ‘head-up, eyes-out’ ego-centric interface philosophy. The cueing systems sought to incorporate common pilot mission task elements into the design. Changes to both the path preview and trajectory prediction were studied. The visual elements of the cues were displayed as if they were comprised of three dimensional physical objects. Operational flexibility in high workload environments is key to pilot task accomplishment. The ability to dynamically generate on demand flight trajectories that pilots could manually fly was another goal of this dissertation. The mathematical framework of Bézier curves was utilized for trajectory planning to ensure the paths satisfy the needs of the pilot, the certification authorities, and the specific mission task element. Four different cueing paradigms were programmed into the Georgia Tech reconfigurable rotorcraft flight simulator. These paradigms were; a 2D Head Up Display (HUD), a Flight Lead Cueing System (FLCS), a Tunnel In the Sky (TIS), and a 3D Flight Path Marker (FPM). The cues were then evaluated using objective measures and pilot workload surveys in a series of Pilot-in-the-Loop (PIL) studies. A total of twenty pilots took part in the study. Seven pilots participated in phase 1, three in phase 2, and ten in phase 4. Phase 3 included only data flown by the author and LTC Joe Davis due to pandemic related travel restrictions preventing the use of additional external pilots. Most PIL studies have a relatively low number of participants, in the range of two to six. In order to gain statistical significance from a relatively low number of participants the participants are asked to repeat the task several times. For example the pilots in phase 4 each flew a total of 54 approaches. The central limit theorem, states that a distribution will be approximately normally for large sample sizes, where a sample size over 30 is considered large. Consequently even when the data is divided to look at a specific cueing condition or starting location the large sample size criteria is met and we can gain statistical insight. Bézier curves provide a feasible method to dynamically generate landing trajectories for pilots to fly by hand. The methods are numerically stable and execute fast enough that there is minimal perceptual latency to the pilot. The pilots were able to follow the generated trajectories with sufficient accuracy both laterally and vertically. The paradigm shift of using 3D AR cueing which the pilots mentally process as signals instead of signs or symbols resulted in reduced workload and had performance that was the same or better than traditional cueing methods.
Computational simulation of adaptation of work strategies in human-robot teams

(Georgia Institute of Technology, 2019-07-22) IJtsma, Martijn

Human-robot teams operating in complex work domains, such as space operations, need to adapt to maintain performance under a wide variety of work conditions. This thesis argues that from the start team design needs to establish team structures that allow flexibility in strategies for conducting the team’s collective work. In addition, team design needs to facilitate fluent coordination of work, fostering the interweaving of team members’ dependent actions in ways that accounts for the dynamic characteristics of the work and the work environment. This thesis establishes a methodology to analyze a team’s strategies based on computational modeling of a team’s collective work, including the teamwork required to coordinate dependent work between multiple team members. This approach consists of the systematic identification of feasible work strategies and the simulation of work models to address the dynamic and emergent nature of a team’s work. It provides a formative analysis tool to help designers predict and understand the effects of their design choices on a team’s feasible work strategies. Two case studies on space operations demonstrate how this approach can predict how work allocation and human-robot interaction modes can foster and/or limit the availability of appropriate work strategies.
Humans teaching intelligent agents with verbal instruction

(Georgia Institute of Technology, 2019-04-15) Krening, Samantha

The widespread integration of robotics into everyday life requires significant improvement in the underlying machine learning (ML) agents to make them more accessible, customizable, and intuitive for ordinary individuals to interact with. As part of a larger field of interactive machine learning (IML), this dissertation aims to create intelligent agents that can easily be taught by individuals with no specialized training, using an intuitive teaching method such as critique, demonstrations, or explanations. It is imperative for researchers to be aware of how design decisions affect the human’s experience because individuals who experience frustration while interacting with a robot are unlikely to continue or repeat the interaction in the future. Instead of asking how to train a person to use software, this research asks how to design software agents so they can be easily trained by people. When creating a robotic system, designers must make numerous decisions concerning the mobility, morphology, intelligence, and interaction of the robot. This dissertation focuses on the design of the interaction between a human and intelligent agent, specifically an agent that learns from a human’s verbal instructions. Most research concerning interaction algorithms aims to improve the traditional ML metrics of the agent, such as cumulative reward and training time, while neglecting the human experience. My work demonstrates that decisions made during the design of interaction algorithms impact the human’s satisfaction with the ML agent. I propose a series of design recommendations that researchers should consider when creating IML algorithms. This dissertation makes the following contributions to the field of Interactive Machine Learning: (1) design recommendations for IML algorithms to allow researchers to create algorithms with a positive human-agent interaction; (2) two new IML algorithms to foster a pleasant user-experience; (3) a 3-step design and verification process for IML algorithms using human factors; and (4) new methods for the application of NLP tools to IML.
Furthering human-robot teaming, interaction, and metrics through computational methods and analysis

(Georgia Institute of Technology, 2019-03-29) Ma, Mingyue (Lanssie)

Human-robot teaming is a complex design trade space with dynamic aspects and particulars. In order to support future day human-robot teams and scenarios, we need to assist team designers and evaluators in understanding core teaming components. This work is centered around teams that complete space missions and operations. The central scope and theme of this work target the way users should design, evaluate, and think about human-robot teams. This work attempts to do so by defining a framework, conceptual methodology, and operationalized metrics for human-robot teams. We begin by scoping and distilling common components from human-only teaming and human-robot teaming research based in areas such as human factors, cognitive psychology, robotics, and human-robot interaction. Taking these constructs, we derive a framework that describes and organizes the factors, as well as relationships between them. This work also presents a theoretical methodology to support designers to understand the impact teaming components have on expected interaction. This methodology is implemented for four case studies of distinct team types and scenarios including moving furniture, a SWAT team operation, a rover recon, and an in-orbit maintenance mission. After assessing various existing methodologies and perspectives, we derive metrics operationalized from work allocation. To test these learnings, this work modeled and simulated human-robot teams in action, specifically in an in-orbit maintenance scenario. In addition to analyzing simulation results given different team configurations, task allocations, and teamwork modes, a HITL experiment confirmed a human perspective of robotic team members. This experiment also refines the modeling of teams and validates our performance metrics. This dissertation makes the following contributions to the field of human-robot teaming and interaction: 1) Created a new comprehensive framework for human-robot teaming by combining key components of team design and interaction, 2) Developed a method to identify distinct archetypes of interaction in human-robot teams (and showed how they fit into a universal framework), 3) Derived metrics from the HRT framework to capture the teaming elements beyond performance and efficiency; operationalized the method and metrics in a computational framework for simulation and analysis, 4) Extended existing computational framework for function allocation to include the metrics, 5) Demonstrated the sensitivity of effective teams to attributes of both teamwork and taskwork.
Decision support system development for human extravehicular activity

(Georgia Institute of Technology, 2017-08-24) Miller, Matthew James

Human spaceflight is arguably one of mankind's most challenging engineering feats, requiring carefully crafted synergy between human and technological capabilities. One critical component of human spaceflight pertains to the activity conducted outside the safe confines of the spacecraft, known as Extravehicular Activity (EVA). Successful execution of EVAs requires significant effort and real-time communication between astronauts who perform the EVA and the ground personnel who provide real-time support. As NASA extends human presence into deep space, the time delay associated with communication between the flight crew and Earth-bound support crew will cause a shift from real-time to delayed communication. A decision support system (DSS) is one possible solution to enhance astronauts’ capability to identify, diagnose, and recover from time critical irregularities during EVAs without relying on real-time ground support. The contributions of this thesis are two fold. The first is domain specific and addresses the known deficiencies that will impact future human EVA operations. The second is methodological and generalizable across many domains. This thesis demonstrates that Cognitive Work Analysis (CWA) can be applied to yield design insight in the form of high level design requirements amenable to traditional systems engineering. Beginning with the first two phases of CWA, a broad work domain analysis of EVA is made to identify the system constraints on EVA operations. Subsequently, Control Task Analysis models were developed that yielded a set of DSS design requirements in the form of cognitive work and information relationship requirements which reflect the underlying states of knowledge associated with supporting EVA operations. Furthermore, this thesis demonstrates how a subset of those requirements, along side envisioning and testing within a future work context, can yield prototype DSS designs suitable for supporting future EVA operations. Finally, this thesis included a human-subject study to evaluate the resultant prototypes against the requirements to demonstrate both validity of the requirements and the verification of the design. As a result, this thesis contributes the underlying science needed to design a DSS within the EVA work domain for future mission operations.
Decision making with incomplete information

(Georgia Institute of Technology, 2017-05-09) Canellas, Marc Christopher

Decision makers are continuously required to make choices in environments with incomplete information. This dissertation sought to understand and, ultimately, support the wide range of decision making strategies used in environments with incomplete information. The results showed that the standard measure of incomplete information as total information, is insufficient for understanding and supporting decision makers faced with incomplete information. The distribution of information was shown to often be a more important determinant of decision making performance. Two new measures of the distribution of incomplete information were introduced (option imbalance and cue balance) and tested across three computer simulations of 18 variations of decision making strategies within hundreds of environments and millions of decision tasks with incomplete information, and one human-subjects study. The simulations were powered by a new general linear model of decision making which can efficiently and transparently model a wide range of strategies beyond the traditional set in the literature. Of the many potential mediators of the relationship between the distributions of incomplete information and performance, only the strategies' estimates of missing information were significant in the computational studies. Accurate estimates resulted in total information being the only meaningful determinant of accuracy while inaccurate estimates resulted in low option imbalance and high cue balance causing high accuracy. The simulation results were partially contradicted by a study in which human decision makers with accurate estimates were affected by option imbalance and cue balance in the same manner as inaccurate estimates – suggesting that some distributions might simply be difficult regardless of the estimates. These results argued that decision support should modify the presentation of information away from difficult distributions. These arguments were codified as heuristic information acquisition and restriction rules which, when tested, increased accuracy without probability and cue weight information.
System design considerations for human-automation function allocation during lunar landing

(Georgia Institute of Technology, 2013-07-08) Chua, Zarrin K.

A desire to advance humanity's presence in space prompts the need for improved technology to send crew to places such as the Moon, Mars, and nearby asteroids. Safely placing a crewed vehicle on and in any landing condition requires a design decision regarding the distribution of responsibilities between the crew and automation. In this thesis, a cognitive process model is used to determine the necessary automated functionality to support astronaut decision making. Current literature lacks sufficient detailed knowledge regarding astronaut decision making during this task and observations of astronauts landing on the Moon are not readily available. Therefore, a series of human-in-the-loop experiments, one of which was conducted with the NASA Astronaut Office at Johnson Space Center, have been conducted to examine the changes in performance due to differing function allocations, trajectory profiles, and scenario operations. The data collected in the human-in-the-loop study has provided empirical data that has informed the cognitive process model, the requirements analysis, and provided insight regarding cockpit display usage and information needs. The proposed system requirements include design guidance for assisting astronauts during both nominal and off-nominal landing scenarios.
Airspace analysis and design by data aggregation and lean model synthesis

(Georgia Institute of Technology, 2013-06-26) Popescu, Vlad M.

Air traffic demand is growing. New methods of airspace design are required that can enable new designs, do not depend on current operations, and can also support quantifiable performance goals. The main goal of this thesis is to develop methods to model inherent safety and control cost so that these can be included as principal objectives of airspace design, in support of prior work which examines capacity. The first contribution of the thesis is to demonstrate two applications of airspace analysis and design: assessing the inherent safety and control cost of the airspace. Two results are shown, a model which estimates control cost depending on autonomy allocation and traffic volume, and the characterization of inherent safety conditions which prevent unsafe trajectories. The effects of autonomy ratio and traffic volume on control cost emerge from a Monte Carlo simulation of air traffic in an airspace sector. A maximum likelihood estimation identifies the Poisson process to be the best stochastic model for control cost. Recommendations are made to support control-cost-centered airspace design. A novel method to reliably generate collision avoidance advisories, in piloted simulations, by the widely-used Traffic Alert and Collision Avoidance System (TCAS) is used to construct unsafe trajectory clusters. Results show that the inherent safety of routes can be characterized, determined, and predicted by relatively simple convex polyhedra (albeit multi-dimensional and involving spatial and kinematic information). Results also provide direct trade-off relations between spatial and kinematic constraints on route geometries that preserve safety. Accounting for these clusters thus supports safety-centered airspace design. The second contribution of the thesis is a general methodology that generalizes unifying principles from these two demonstrations. The proposed methodology has three steps: aggregate data, synthesize lean model, and guide design. The use of lean models is a result of a natural flowdown from the airspace view to the requirements. The scope of the lean model is situated at a level of granularity that identifies the macroscopic effects of operational changes on the strategic level. The lean model technique maps low-level changes to high-level properties and provides predictive results. The use of lean models allows the mapping of design variables (route geometry, autonomy allocation) to design evaluation metrics (inherent safety, control cost).