Overview
Automation in Procedures: Guidelines for Allocating Tasks for Performance
Principal Investigator:
Debra Schreckenghost, M.E.E.
Organization:
TRACLabs Inc.
Technical Summary
As crewed missions move deeper into space and communication latency increases, strategies for carrying out tasks must shift. Astronauts will be unable to depend on real-time support from flight controllers; controllers will not be able to perform procedures in the same way they do for the International Space Station, nor to advise on changed applicability of procedures in real-time. This change threatens to increase astronaut workload, decrease efficiency and increase the risk of suboptimal task execution.
Automation is an important resource for adapting to this altered environment. To help rather than harm, however, automation must be effectively integrated with the humans it supports.
Specific Aims
1) Identify and refine candidate strategies for allocating tasks to automation and the factors when guidelines apply;
2) Define effectiveness measures for these task allocation strategies;
3) Conduct empirical assessment of the effectiveness of human-automation integration based on the proposed allocation strategies, and
4) Integrate the project’s findings as proposed task allocation strategies with automation.
The researchers’ approach to automation investigates the use of procedures as the basis of automation. By using a human-oriented procedure to organize automation, automation is designed to be more comprehensible to operators. For procedure automation, the actions in a procedure are enabled for automatic execution when instrumentation is available to perform the action. The degree of automation can be varied by changing which actions are designated for automatic execution. Strategies for determining which steps should be automated range from pre-defined allocations to flexible adjustment while the procedure is being executed. It is not clear how these strategies affect human-automation performance. Research is needed to determine how these different allocation strategies affect task performance and learnability. Additionally it is not clear how robust these strategies are to changes in situation that invalidate the procedure as written.
The researchers seek the identification of several candidate guidelines and factors that define effective task allocation strategies, within a procedure-automation approach. These guidelines consider where benefits may be greatest and how automation can be structured to realize the potential benefits. However, the ability to automate will depend upon reliably determining whether a procedure should be applied exactly in the circumstances and, if not, how to provide human skills to ensure appropriate application. The way in which human and automated actions are coordinated also needs to make the work organization meaningful to humans, as well as reliably executable by the automation. Specifying how to identify meaningful units of work that serve as the foundation for coordination between human and automated actions is a core part of the research.
TRACLabs and its partners from NASA and San Jose State University will evaluate human performance for different task allocation strategies for procedure automation and use the results to articulate a set of allocation strategies. They will define operational scenarios for the evaluation, including a set of multi-step procedures and simulation that works with these procedures. They will provide procedure automation software for executing these procedures on the simulation. The researchers will perform ground-based human subject testing where subjects use the procedure automation to perform the procedures and will select the task allocation strategies for evaluation that include both predefined and flexible allocation of tasks. They will measure and analyze human and automation performance for each of these strategies under both nominal and off-nominal circumstances. Finally, the researchers will use experimental results to derive strategies for task allocation that are an important step toward developing technology to guide the allocation for tasks among humans and automation.