Title: Physical Modeling of Time and Motion Estimation in Manual Assembly


Name: Chao Wang, PhD

(former student)


Project description:


Starting situation

Time and motion study have been used to generate standard times for human motion in the manufacturing industry. These standard times can be used to estimate the total time required to perform various manual work, which helps the design as well as management team to improve the efficiency of an assembly line. The standard time values are compiled into tables for use by analysts who have received training on using the tabulated standard time data. This study aims to develop a new method to estimate movement time for a human performing a pick-and-place task by using a control theoretic model. This new method can enhance the tabulated data by offering estimated movement time values for those not found in the standard time tables.



It was observed that human pilots usually adapts to the vehicle’s dynamics in order to achieve adequate control performance. The crossover model developed by McRuer and Graham sumarizes the adaptation of pilot by observing that the combined pilot-vehicle system’s frequency response resembles that of a integrator at frequencies near the crossover. The structural model developed by Hess inplements a dual-loop architecture to represent the neuromuscular system of the human arm when performing a target tracking task, and the inner loop is shaped by following the crossover model.

In this study, experimental data is gathered to identify the parameters of the crossover model for a human worker performing two target acquisition tasks: touchscreen pointing and pick-and-place. A structural model is then tuned to mimic the performance of the crossover model by using Quantitative Feedback Theory techniques, which offers a intuitive graphical loop shaping procedure while allowing for uncertainty in the plant.


Expected Results

The output of the tuned model consists of a single axis time-history of the hand position, which compares favorably with another set of experimental data. It is capable of predicting a time-accuracy trade off typically observed during aimed pointing tasks. It also offers more accurate movement time estimation compared to existing standard time tables, since the model is “customized” to the operator performing the experiments.