Semester of Graduation

August 2022


Master of Science in Industrial Engineering (MSIE)


Mechanical and Industrial Engineering

Document Type



Manufacturing robotics have been used for decades to perform repetitive tasks, or tasks that require increased levels of speed, strength, or precision to meet production and specification requirements. Determining the appropriate degree of automation for both the human and robot collaborative team members is critical to optimize production as well as user experience. If the degree of automation is too high, the human will be out of the loop which can result in the loss of situational awareness and be detrimental to intervention time and accuracy. If the degree of automation is too low, then the human may experience greater than necessary cognitive workloads resulting in fatigue.

In this experiment, the human team member performed a typing task while the robot team member performed a pick and place task. The human remained in a supervisory role to the robot’s actions. As the degree of automation increased, objective and subjective measurements of the cognitive effect were collected. 30 participants aged 18-44 years were divided between two levels of automation. Group one (n=15) participated in the decision support variation of the experiment. The human team member typed from provided literature while they supervised the Universal Robot performing a pick and place task. The robot then initiated movement to pick up the object when the human pressed a button. Group two participated in the automatic execution variation of the experiment in which the robot executed all movements automatically, without input from the human operator. The robot performed the pick task when an object was detected by a photelectric switch. In each scenario, an error was presented on the 5th, 9th, 15th, and 19th production cycle requiring the human to intervene. Time measurements were collected at this juncture to determine how the human reacted to an unexpected situation at each level of automation.

The Subjective Workload Assessment Technique (SWAT) (Luximon & Goonetilleke, 2001) was used to measure the perceived cognitive workload for each participant. Time data were collected for the production cycle and intervention. Typed words per minute were collected and compared to each participant’s control. T-tests were used to analyze the mean time within groups comparing the control words per minute to the treatment words per minute. Additionally, t-tests were used to analyze mean time between groups in the following areas, time to complete twenty cycles, intervention time, and words per minute.

Two hypotheses were developed which results were measured against. The first hypothesis is that the automatic execution operators will be out of the loop while engrossed in the typing task and take more time to notice the error, understand the problem, correct the problem, and return the system to homeostasis than decision support operators. The experiment results did not fully support this hypothesis, but they did reveal that the increased level of automation completed 20 cycles significantly (p < 0.0001) faster mean time 9.57 min (SD 0.2 x 10-3) than the lower level of automation with a mean time 10.25 min SD 0.2 x 10-4. The second hypothesis is that automatic execution operators will be faster and more accurate in word processing and have lower subjective workload ratings than decision support operators. This hypothesis was not supported in the experiment results which revealed nearly significant (p = 0.07) higher levels of cognitive effort in higher levels of automation (57.5%) than in the lower level of automation (45.3%) (SD 2.0). However, no significant differences were found for word processing time or errors. The results of this study suggest that as automation increases overall production outcomes increase, but so does mental workload. This may help researchers and automation designers understand the relationship between cognitive workload, performance, and levels of automation within a human robotic collaborative team (HRC).

Committee Chair

Ikuma, Laura