RoBo Diagnostics

Interactive software app to help communicate with welding hardware devices and increase efficiency in the repair welding process.

Timeline: 4 months Role: User research, Product design, Prototyping

Role

UX/UI Designer
(Project Member)

Team

Iowa State’s HCI Class
(2 engneers 2 HCI students)

Duration

Sept. 2023
(4 mo. total)

The Process

— 1

Discovery

— 2

Ideation

— 3

Design

— 4

Test

User Interview & Onsite Discovery

The study found that the production welding operators take a lot of pride in their ability to use their expertise to identify anomalies in the casting, determine a remediation process, and sign off on the quality of the remediation. However, these users did not enjoy the repetitive aspects of the process. An operator made a tongue-in-cheek joke that “It's supposed to be a monotonous job”.

These findings led the research team to focus development on a system that could reduce the repetitive work and remove the operators from the harsh conditions while increasing the time they spend making expert judgments and decisions.

01

Operators take a lot of pride in expertise work.

03

Harsh working inviorment.

02

Data Interoperation & Results

During the interview and observation processes, notes were recorded to document the data that was gathered. This data was grouped based on themes by the research team members. The result of this interpretation process was an affinity diagram. 

The themes identified during the creation of an affinity diagram were used to form personas for each of the user groups and to conduct a hierarchical task analysis (HTA). These tools can then be used to identify areas where automation can help the users and where user interface needs exist for the user to interact with that automation..

Too many repetitive taks for a skilled job.

What did we learn?

Wireframes & Low Fidelity

After researching, pivoting, and understanding the hierarchy of tasks, our team developed early designs. We began by first drawing out our assumptions.

⸻ TESTING

Identifying Usability issues

Cognitive Walkthrough

The Participants 
The participants for our non-user evaluation were members of another group – Team 13 – in our course. Hannah Ragsdale Lee, Javier Ahumada, Jesse Sherrill, and Rachel Schmitz evaluated our design. 

Tasks / Scenarios 
The tasks analyzed in our cognitive walkthrough were as follows: 

  • Goal 1: Scan new casting and view scan progress

  • Goal 2: Evaluate errors, create instructions, and send to robot.

  • Goal 3: Quality check fixed errors and give approval.

Results & Issues addressed 
The data from the evaluation was used by the team to distill down to a core set of usability issues. Each issue will be rated on a severity scale described in Table 1

User Testing

 The objective of this study is to determine if there is value in investing in an intelligent assistant for casting defect classification that could be used to pre-populate certain fields in the user interface. This study aims to evaluate the value of such an assistant by evaluating a simulated version of this intelligent system and determining if it has potential to improve user performance. This study will achieve this objective by testing two main hypotheses. First, this study will test that an intelligent assistant will improve the speed at which a user can complete the assignment of process parameters to the repair process of a casting. Second, the researchers hypothesize that the users will experience an improvement in their evaluation of the interface on the System Usability Scale (SUS). These hypotheses will be tested while monitoring the user performance to ensure that their accuracy does not decrease.

Test Setup

The Participents 

The target participants for this study will be undergraduate and graduate students at Iowa State University. They will be pursuing either a mechanical or industrial engineering degree and will ideally have familiarity with both the welding and metal casting processes. Participants will be recruited from the pool of undergraduate and graduate research assistants working in the Slater Lab for Advanced Manufacturing, but who do not work directly with project tasks related to this study. These participants will have visited a steel foundry and had a firsthand account of the production welding process, making them a better proxy for a real production worker than a typical engineering student. If additional participants are needed, we will request that our current participants recommend other participants and will recruit them through the IMSE (Industrial Manufacturing and Systems Engineering) department. This pilot study intends to evaluate the performance of five participants. While it would be ideal to have more participants to improve the likelihood of establishing statistical significance and to have actual production welders as participants, the researchers believe that this limited study will provide the team with preliminary data that could help justify a more in-depth study in the future. 

Tasks / Scenarios 

The participants will complete the user goal two defined in previous reports, which consists of the user evaluating the errors, creating instructions for the robot, and sending those instructions to the robot. The users will be guided by the statement “Now that you have diagnosed and identified the errors in the casting, review the three errors that were identified and begin creating instructions for the robot to repair each defect.” The user will then walk through the screens of the interface to complete this task both with and without the intelligent assistant. 

Methods

Dependent Variables 

The dependent variables measured in this evaluation include usability, task completion, and task performance. To measure the usability of our interface participants will fill out a System Usability Scale (SUS) questionnaire after each trial. The SUS questionnaire will be made up of 10 questions that participants will answer based on a Likert Scale from 1 to 5 with 1 meaning strongly disagree and 5 meaning strongly agree. 

  • User Satisfaction

  • Task Completion

  • Task Performance

Independent Variables 

The independent variable for this experiment will be the process of data input. Participants will be using two different interfaces to complete the same task. Each participant will be choosing repair instructions for multiple cast errors using a level 1 and level 2 interface. The repair instructions include a series of input fields that users must fill in to proceed. The level 1 interface will have input fields that users must fill in one-by-one, while the level 2 interface will have input fields automatically filled in. 

  • Experiment Observation Type: Direct Observation

  • Experiment Location: Controlled Usability Lab

  • Experiment Type: Within Subjects

Building structure

As the saying goes, “A good start is half the battle." Before going into user interface design, we made sure to polish the features and user interaction flow.

In our future application interface, users will be able to conduct the very manual process of repairing errors in cast components in a very automated way. Automating this process while still giving users oversight was the main goal of our future application. Identifying, analyzing, and repairing errors is done by hand today, but with our application and in conjunction with advanced robotics this process will potentially be quicker and more efficient.

⸻ DESIGN DIRECTION