copy and redistribute the material in any medium or for

Despite the recent advances in automation, the role of humans in the HCS is still considered a key factor for adaptability and flexibility. We therefore propose a method to quantify task complexity for effective management of the semi-automatic systems such as an MMAL.

Introduction

• Background
• Motivation
• Objectives
• Affordance-based choice modeling
• Computation of choice complexity
• Human in the loop simulation incorporating choice complexity
• Problem Statement
• Assumptions and Research Scope
• Research Questions
• Research Strategies
• Dissertation Overview

Recall that increased human flexibility leads to a "complexity of choice," with a wide range of implications. Thus, based on the information, this research proposes a formal quantification of choice complexity in human-centered system such as mixed model assembly line.

Weighted Affordance Based Task Modeling in the Simulation of Human Centered System

Introduction

The action selection process of human agents, i.e. the initiation of state transitions, is modeled stochastically in accordance with the action-state cost (load) values. This article aims to assist system designers of the safety-critical systems by providing a systematic understanding of the interaction between heterogeneous agents and dynamic environments, with the implication of the agent-dependent decision characteristics.

Background

• Affordance theory and prospective control
• Human decision models

In this study, we add two more terminologies regarding the temporal-spatial availability of accessibility: “expected (or imagined) accessibility” and “perceived (or real) accessibility. In the proposed model, we integrate accessibility theory into human agent modeling to capture the dynamic agent-environment interaction.

Figure 2-1. Affordance and perceptual information [14]

Weighted Affordance-based Action

• Weighted affordance of agents
• Agent decision making process

Also, using 𝑥𝑟𝑠(𝑎) ∘ 𝑣𝑖( To incorporate cost into the agent's decision making, a Markov decision model is used in which agent 𝑖 incurs cost 𝑐𝑖(𝑠, 𝑎, 𝑠′) for performing action 𝑎 and reaching state 𝑠′ from state 𝑠.

Figure 2-3. Action Selection Framework Attributes:

Building Evacuation Simulation

• Policy
• Simulation Results
• Parameter Overview
• Results

NASA TLX distribution for Frustration (F) for four actions when performed on the 4th floor. For example, in the simulation, the load estimate for agents located on the sixth floor in a secure environment follows a truncated normal distribution with the mean and standard deviation shown in Table 2-4. NASA-TLX rating for four actions performed on the sixth floor in a secure environment.

Interaction means that the influence of the factor on the agent's behavior also depends on the level of the other factor.

Figure 2-5. The “take-a-bility” affordance is only available when the elevator is in “Not full” state

Conclusion

In addition to the general framework, an evacuation simulation study was performed as an example to verify the feasibility of the proposed simulation framework. The evacuation example shows how the proposed framework can accommodate the MDP and the perception of advice in the planning and execution levels. Due to the subjective nature of human subjects and the small size of human subjects involved in estimating NASA-TLX data, the simulation results cannot be generalized; instead, the example should serve as an exemplary template for the proposed simulation framework.

Despite the limitations, we anticipate that the proposed model can be used to examine problems of human involvement in system design.

Computational modeling of task complexity in human-centered systems: A Case of a Mixed

Introduction

In the automotive industry, analyzing manufacturing complexity is a reasonable way to ensure higher product variability while maintaining production efficiency. In addition, these conflicting aspects of complexity in the system have additional direct and/or indirect costs for managing the manufacturing process and associated resources [4]. Thus, a cost-benefit scenario is typically studied to justify the variety introduced into the manufacturing line.

Despite a recent advance in manufacturing automation, the role of humans in the manufacturing system is still considered a key factor in adaptable and flexible systems, such as in a mixed-model assembly line.

Literature Review

Hick's law is colloquially used to describe the time it takes for someone to make a decision as a result of the possible choice [9]. Although the similarity of options in a mixed-model assembly line clearly affects the choice complexity of the system operations and is detrimental to the overall system performance, the formulation and analysis of the similarity of options has received less attention from researchers and practitioners in the manufacturing industry . Assuming proximity, the selection of a target drug among several similarly named drugs has been shown to increase the difficulty of visual search for the target.

Similar conclusions have been observed to some extent regarding the shape differentiation of the drugs.

Choice Complexity Model

• Similarity Measure in Semantics
• Similarity in a Mixed-Model Assembly Line
• Incorporating Similarity Measure in a Complexity Model

One of the well-known similarity measures built on the basis of the commonalities and differences between two semantic representations is the feature-based similarity measure. Note, however, how the similarity measure strongly depends on the representation of the discriminative features, regardless of the complexity of the part. That is, Θ maps a given target variant to the sum of the pairwise similarities between the target variant and each alternative option as shown in Equation (3-5).

According to equation (3-7), 𝑣𝑖𝑡𝑘 is the target option, and the overall level of activated similarity is simply the sum of the pairwise similarities between 𝑣𝑖𝑡𝑘 and each other available option, which is equivalent to the sum of the row 𝑡𝑡ℎ of the similarity matrix S .

Figure 3-1. (a) Feature-based similarity measuring model in set-theoretic operations, (b) Object representation via an unstructured set of features

Illustrative Case Study

• Screw Choice Complexity
• Choice Complexity and Reaction Time
• An Experimentation Overview
• Feature Selection and Complexity Measure
• Result and Discussion

This is because the similarity of options is one of the underlying factors of choice complexity. Given an increase in the demand share for option A, for example, its contribution to the level of choice complexity varies as shown in Figure 3-4(b). An example of the stimulus and the pool of options in the form of Lego pictures is shown in Figures 3-5.

In the above experiment, the discriminative features chosen in the study are the length and color of the Lego pictures.

Figure 3-2 A set of screws and their variety

Conclusion

Thus, it is possible to determine the parts associated with the highest level of complexity and integrate them into the system with the help of the proposed model. In addition to calculating the complexity of the system, it can be shown that with a fixed number of options, both the level of similarity and closeness can be adjusted to reduce the complexity of the system. On the one hand, if customers base their purchasing decisions on specific features, increased complexity can be offset by increased sales.

On the other hand, additional costs associated with the increased complexity may be unjustified with lower demand.

Modeling framework for Human in the Loop Simulation of Task Complexity: A Case of

Introduction

58 As in most complex systems in which closed-form analytical solution does not exist, simulation has become a powerful tool in the analysis of complex manufacturing systems. Thanks to the technological advances in the new "smart manufacturing" era, simulation analysis has made its stride. This chapter aims to simulate and analyze the OCC and its underlying effects, by including a real human (human-in-the-loop) in the overall assembly simulation in a way that captures the core physical aspect of accurately represent choice complexity.

The proposed methodology provides an in-depth analysis of OCC and gives a suggestion on how to effectively summarize the human component in intelligent manufacturing environments.

Literature review

• Manufacturing Complexity
• Modeling and Simulation of Human-involvement in Manufacturing Systems

Next, Section 4 provides an in-depth discussion on the features and characteristics of choice complexity and its impact in MMAL. The success of the simulation is based on the progress in the representation of several aspects of the manufacture in computable terms (ie conceptual model)[93]. While several aspects of manufacturing complexity can be modeled and analyzed, choice complexity presents a greater challenge due to the multiple human factors involved.

Due to the complexity of human actions, the existing human models are often oversimplified and only built for specific purposes (eg military, etc.).

Figure 4-1 Screw Selection task. The subject clicks on a specific screw according to the stimulus

Human in the loop machine learning

• Machine learning in manufacturing systems
• HIL machine learning in MMAL

While most virtual simulations focus on practice and testing user knowledge using scenarios and interactive environments to reflect real-life situations[98]; in this study, we propose machine learning-based HIL simulation in which a non-immersive virtual picking simulation is used to accurately represent and analyze OCC and predict its impact on manufacturing performance. Machine learning has been used successfully in process optimization, monitoring, control applications and predictive maintenance in various manufacturing industries. While most applications of the machine learning concept have often been limited to the optimization of scheduling or line balancing problems, machine learning has, at times, been applied to predict the task duration of a wide range of manufacturing processes.

After selecting the choice complexity features, we build and train a machine learning model based on the operator's choice time as depicted in Figure 4-3, which will be further discussed in the next sections.

Figure 4-3 Human in the loop machine learning in a MMAL.

Features and characteristics of choice complexity in MMAL

• Simulation of choice complexity in MMAL
• Features of operator’s choice complexity
• Similarity of variants
• Sequence Rule
• Physical arrangement
• Result and discussion
• Machine learning and prediction of selection time
• Human-involved system control and simulation

In the next section, we will discuss three of the more complex selected features of choice complexity in a mixed model assembly line and the form in which they are assembled in this illustrative example. A perfect prediction of the operator's selection time is unattainable; however, the proposed method reasonably mimics the actual physical setups that define the choice complexity in a mixed model. Changes in the mean of the actual reaction time (RT) caused by various parameter changes (per random forest algorithm).

As the number of alternative choices grows, changes in choice time strongly depend on the sequence rule.

Figure 4-5 Selected features

Conclusion

Human involvement in the loop also provides more scope for testing and optimizing the number of assembly line policies. That is, the impending adoption of smart manufacturing in the future will give rise to several possible applications of machine learning in a wide range of human-involved manufacturing processes in which the proposed framework can be applied accordingly. Despite the limitations, the proposed model is a valuable tool in the pursuit of an effective modeling and simulation of human-centered complex systems.

Thus, the proposed model is an example of how we can effectively integrate the human component into smart production environments.

Conclusion

Research Summary

The proposed model not only has the ability to calculate the overall complexity of the system, but can also trace the contribution of each specific option or station to the overall system complexity. We conducted a simple experimental design to verify the effect of the similarity of options and reaction time on the overall complexity. Thus, the fourth chapter proposes a HIL simulation framework where different parameters of choice complexity are tested to assess the overall effect on operators' efficiency.

The proposed model, together with an illustrative case study, not only serves as a tool to quantitatively assess the impact of choice complexity on operator effectiveness, but also provides insight into how to reduce complexity without affecting overall production throughput. to influence.

Research Contributions

In HCS such as MMAL, human performance, a key factor in adaptability and flexibility, is of great interest due to its impact on the overall system. The model can thus be used to assess different scenarios and their respective effects on the overall complexity and reliability of the system. Finally, the proposed HIL simulation framework offers the tool to investigate human performance in HCS such as MMAL.

The proposed HIL framework provides an ability to investigate the impact of human behavior on the performance of the entire HCS, especially the manufacturing system.

Research Limitations and Future works

Butala, “Assessing the operational complexity of manufacturing systems based on statistical complexity,” International Journal of Production Research, vol. Limère, “Workload balancing and production complexity leveling in mixed-model assembly lines,” International Journal of Production Research, pp. Marin, “Modelling Production Complexity in Mixed Model Assembly Lines,” Journal of Manufacturing Science and Engineering, vol.

Kim, "Computational modeling of manufacturing choice complexity in a mixed model assembly line," International Journal of Production Research, pp.