Real-time scheduling optimization using algorithmic planning and workforce analytics

Most complex planning and scheduling problems are defined by a single structure that encapsulates all essential elements, such as resources, timelines, and tasks. The relationships within this structure encode the constraints and costs associated with achieving particular goals. However, in certain domains, “real-time” is a key aspect of a planning or scheduling problem. In this case, task scheduling choices must be made frequently and rapidly in ways that minimize execution costs and the impact of the resulting schedule on future execution. Furthermore, real-time scheduling often takes place in environments characterized by high uncertainty, with both incomplete or imperfect information and rapidly changing dynamics. In combination, these two factors make it impossible to model real-time planning and scheduling problems within a fixed and fully-specified structure. For a team of human or robotic agents executing a real-time operation, the structure itself is typically the result of collaborative execution, continually evolving through both the actions of individual agents and their dynamic responses to the actions of others. Little, if any, pre-specified scheduling infrastructure actively constrains their interactions. Moreover, real-time task allocations yield not only schedule plans but also compelling social cues governing the task-specific behavior of co-actors. For a group of humans, these cueing effects can cause unexpected and unwanted changes in schedule-enforced task performance, as a result of behavioral mores and expectations.