Putting Time (back) into Dynamic Function Allocation

This position paper discusses extensions to concepts of Dynamic Function Allocation (DFA) that could improve our understanding of the trade-offs involved in designing and operating humanautomation systems. We suggest that current DFA paradigms, focusing predominantly on allocation along the human-automation resource dimension, may provide an insufficient basis for design decisions as they fail to take account of alternative function management strategies. Of these strategies, Dynamic Function Scheduling (DFS), the allocation of functions along the temporal dimension, is of particular interest, not least because scheduling is both a mature engineering discipline and a ubiquitous aspect of human behavior. Understanding these scheduling decisions requires consideration of the temporal properties of functions (e.g. continuous, periodic, sporadic, pre-emptable, interleavable), temporal requirements (e.g. deadlines), and the temporal properties of the agents, human or automatic (e.g. service rates, interruption handling, task switch costs, temporal reasoning abilities, control modes). The paper reviews engineering and human factors approaches that could support the representation, analysis and design of DFS.