Todd J. Callantine

Agents for analysis and design of complex systems

The paper describes how intelligent agents can simulate human operators to aid in the analysis and design of complex systems. The paper presents two examples of adapting the Crew Activity Tracking System (CATS) to function as an intelligent agent. The first is a model-based design application in which CATS agents perform the task of air traffic controllers in order to test a new operational concept. The second concerns human error analysis, in which a coordinated team of CATS agents represents a flight crew and aircraft destroyed due to a controlled flight into terrain. The paper also discusses issues regarding the development of task model-based agents that make plausible, human-like errors and use such agents.

NASA's ATM technology demonstration 1 (ATD-1): Integrated concept of arrival operations

This paper describes operations and procedures envisioned for NASAs Air Traffic Management (ATM) Technology Demonstration #1 (ATD-1). The ATD-1 Concept of Operations (ConOps) demonstration will integrate three NASA technologies to achieve high throughput, fuel-efficient arrival operations into busy terminal airspace. They are Traffic Management Advisor with Terminal Metering (TMA-TM) for precise time-based schedules to the runway and points within the terminal area, Controller-Managed Spacing (CMS) decision support tools for terminal controllers to better manage aircraft delay using speed control, and Flight deck Interval Management (FIM) avionics and flight crew procedures to conduct airborne spacing operations. The ATD-1 concept provides de-conflicted and efficient operations of multiple arrival streams of aircraft, passing through multiple merge points, from top-of-descent (TOD) to touchdown. It also enables aircraft to conduct Optimized Profile Descents (OPDs) from en route altitude to the runway, using primarily speed control to maintain separation and schedule. The ATD-1 project is currently addressing the challenges of integrating the three technologies, and implantation into an operational environment. Goals of the ATD-1 demonstration include increasing the throughput of high-density airports, reducing controller workload, increasing efficiency of arrival operations and the frequency of trajectory-based operations, and promoting aircraft ADS-B equipage.

A representation of air traffic control clearance constraints for intelligent agents

This paper presents an object-oriented representation of environmental constraints that are conveyed by air traffic control (ATC) clearances. It presents background research, describes the representation, and discusses how it enhances context representation in an intelligent agent called the Crew Activity Tracking System (CATS). The paper also discusses implications of such a representation for related intelligent agent research in the aviation domain.

Model-based crew activity tracking for precision descent procedure refinement

New procedures for human operators in complex systems must be both operationally effective and inexpensive to introduce. Procedures may require refinement to meet these objectives. The paper illustrates how process measures about operator performance using a new procedure can be derived via activity tracking, and how this information can supplement product measures to support procedure refinement. Activity tracking is a form of intent inferencing suitable for constructing and maintaining a representation of operator activities in real time (T.J. Callantine et al., 1997). The Crew Activity Tracking System (CATS) uses a normative model of the new procedure to predict how operators should perform the procedure; operator actions that CATS cannot interpret signal potential operator difficulties. By using the model to track the activities of operators using the new procedure, CATS can identify mismatches between the model and actual operator actions. Through this process, CATS can help substantiate potentially problematic aspects of the procedure. Difficulties identified by CATS then serve as the focus for further refinements to the procedural elements, or the supporting documentation and training used to introduce it.

Computational modeling of air traffic control: terminal area case study

This paper presents an air traffic control modeling case study in which agent performance on a simulated terminal-area air traffic control task was compared with human air traffic controller performance. The paper first provides background on the simulation with human air traffic controllers, and on prior air traffic control modeling research. It then presents an agent model that enables various air traffic control strategies to be evaluated. Results for four different control strategy models show that agent performance compares more favorably to human performance when clearance alternatives and resultant aircraft behaviors are more constrained. The paper discusses these results and describes proposed future enhancements. The research was conducted with support from the NASA Aviation Safety Program.

Detecting and Simulating Pilot Errors for Safety Enhancement

This paper presents research on computational models of pilot activities for detecting and simulating pilot errors, and discusses how these techniques may be used to enhance aviation safety. The discussion addresses improved feedback about pilot errors to support training, and the use of actual and simulated error data to understand how errors may impact new air traffic management concepts and flight deck automation. The research is supported by the System-Wide Accident Prevention project of the NASA Aviation Safety Program.

Analyzing constraints to support computational modeling of air traffic controllers

Research toward enhancing computational models of air traffic controllers for use in assessing new air traffic management concepts is presented. Simulation data from professional air traffic controllers is analyzed using a computational method for characterizing the constraints in force when the controllers take action. The results show controllers apply strategies to proactively prevent losses of separation between aircraft. This approach has advantages for complexity reduction and workload management and implications for human roles in future air traffic management systems.

Context in Models of Human-Machine Systems

All human-machine systems models represent context. This paper proposes a theory of context through which models may be usefully related and integrated for design. The paper presents examples of context representation in various models, describes an application to developing models for the Crew Activity Tracking System (CATS), and advances context as a foundation for integrated design of complex dynamic systems.

Air and Ground Simulation of Terminal-Area Traffic Management with Airborne Spacing

Controller and pilot decision support tools for operations with airborne spacing in the terminal area were evaluated in a simulation conducted at NASA Ames Research Center as part of the NASA Advanced Air Transportation Technologies project Distributed Air Ground Traffic Management element. The results indicate that airborne spacing improves spacing accuracy and may help reduce go-arounds. Controller workload is acceptable and spacing clearances containing lead aircraft callsigns are clear. Expected operational benefits depend on traffic flow coordination and predictable spacing guidance and support tool behavior.

Model-based design of air traffic controller-automation interaction

A model of controller and automation activities was used to design the controller-automation interactions necessary to implement a new terminal area air traffic management concept. The model was then used to design a controller interface that provides the requisite information and functionality. Using data from a preliminary study, the crew activity tracking system was used to help validate the model as a computational tool for describing controller performance.