MODELING OF TASK COMPLEXITY IN HUMAN-CENTERED SYSTEMS

Abstract

Throughout the years, technological expansion has been coupled with complex work allocation in Human-Centered System (HCS). In spite of the recent advances in automation, role of humans in the HCS is still regarded a key factor for adaptability and flexibility. Meanwhile, due to advances in computing, computer simulations have been the indispensable tool in the study of complex systems. However, due to the inability to accurately represent human dynamic behavior, the majority of HCS simulations have often failed to meet expectations. The failure of HCS simulations can be traced in poor or inaccurate representation of key aspect of system. Whereas the machine component of HCS is often accurately simulated, research claims that human component is often the cause of a large percentage of the disparity between simulation predictions and real-world performance. This dissertation introduces a novel human behavioral modeling framework that systematically simulates human action behavior in HCS. The proposed modeling framework is demonstrated with a case study using simulation in which a set of feasible human actions are generated from the affordance-effectivity duals in a spatial-temporal dimension. The model employs Markov Decision Process (MDP) in which NASA-TLX (Task Load Index) is used as cost estimates. The action selection process of human agents, i.e., triggering of state transitions, is stochastically modeled in accordance with the action-state cost (load) values. A series of affordance-based numerical values are calculated for predicting prospective actions in the system. Finally, an evacuation simulation example based on the proposed model is illustrated to verify the proposed human behavioral modeling framework. The incorporation of human modeling in HCS simulation offers a wide range of benefits in representing human’s goal directed action. However due to the complexity and the cost of representing every aspect of human behavior in computable terms, the proposed framework is better fit in simplified and controllable environment. Thus, we then propose a human in the loop (HIL) approach to investigate the operator’s performance in HCS; particularly, the mixed model assembly line (MMAL). In HCS such as MMAL, human operators are often required to carry out tasks according to instructions. In the proposed methodology, rather than a mathematical representation of human, a real human plays a core role in system operation for the simulation and consequently influences the outcome in such a way that is difficult if not impossible to reproduce via traditional methods. At the initial stage of the simulation, various features are extracted after which, a stepwise feature selection is used to identify the most relevant features affecting human performance. The selected features are in turn used to build a regression model used to generate human performance parameters in the HCS simulation. Finally, we explore the analytical relationship between the flexibility (variation) and the complexity of human role in HCS. As the number of alternative choices (or actions) available to human increases, the choice process becomes complex, rending human modeling and predictability more difficult. The dissertation will particularly utilize the visual choice complexity to convey the proposed computation of task complexity as a function of flexibility. Thus, we propose a method to quantify task complexity for effective management of the semi-automated systems such a MMAL. Based on the concept of information entropy, our model considers both the variety in the system and the similarity among the varieties. The proposed computational model along with an illustrative case study not only serve as a tool to quantitatively assess the impact of the task complexity on the total system performance, but also provide an insight on how the complexity can be mitigated without worsening the flexibility and throughput of the system.