Allocation of Authority and Responsibility in Advanced Air Mobility Operations for Performance-Limited Agents
Open Access
- Author:
- Ghimire, Ishan
- Graduate Program:
- Aerospace Engineering
- Degree:
- Master of Science
- Document Type:
- Master Thesis
- Date of Defense:
- February 23, 2024
- Committee Members:
- Amy Ruth Pritchett, Thesis Advisor/Co-Advisor
Eric Greenwood, Committee Member
Jacob Langelaan, Professor in Charge/Director of Graduate Studies - Keywords:
- AAM
Function Allocations
CONOPS
Monitoring
Working-Memory - Abstract:
- The operation of human-autonomy teams in Advanced Air Mobility (AAM) depends on the capabilities of each agent, what information they need to know and communicate, and the organizational structure that outlines what actions they perform and what outcomes they are responsible for. This thesis examines the Concept of Operations for Urban Air Mobility Maturity Level (UML) 4, hereafter referred to as CONOPS, to highlight the differences and commonalities in two principles for allocating authority and responsibility for all the actions needed to perform the CONOPS: one based by grouping the roles the different agents perform, and one based on the location of where the actions take place. Using Work Models that Compute (WMC), simulations of a case study involving multiple vehicles performing multiple missions examined the demands placed on the agents with each allocation. Additionally, to represent the behavior of human agents more accurately within the teams, this thesis models agents as having limits. Some simulations examined the impact of limiting all agents in the number of actions they can simultaneously perform. In other simulations the human agents were represented as memory limited agents, with a maximum constant value on the number of discrete pieces of information that can be held in their working memory. Since all actions do not require the same amount of information to be held in short-term memory, this creates a dynamic limit to the number of actions simultaneously performed. In considering the agents’ taskload, there is a strong need to also model the monitoring actions done by the agents, where one agent is responsible for the outcomes of the actions executed by a different agent. This monitoring work significantly impacts the load on the agent, resulting in acute and chronic saturation of the agents’ capabilities, and delays in safety critical monitoring.