Cognite summer intern project
Creating a data visualization mini-app
My summer in Cognite started off with one fundamental question: what am I actually doing here?
I was on a project team with 4 other interns and a project lead from Cognite (who were all developers, bar one project management intern). This team was a subset of the in house "robotics" team at Cognite, and our task for the summer was to get a third party robot up and running, collecting data at our client's industrial facility. Part of this included defining what types of data could be useful, and how to make the data useful if processing was necessary.
The plan was for me to work on designing a dashboard that would display the current status of the robot, as well as data that had been collected. The project was organized into weekly sprints, with two sessions (at the beginning and middle of the summer) defining overarching goals.
At the beginning of the project, most of my time was spent gaining an understanding of how the facility we were using as a test case operates, as well as learning about the robot we were using, which use cases we were aiming to achieve, and talking to the operators at the facility about their job. In particular, I was curious about what routine operations could be automated by an autonomous robot, as well as how the information from those tasks is communicated to others in the organization.
After getting a basic idea of what the challenges and opportunities were, I had several whiteboarding sessions (orangeboarding? see above) attempting to map out the information that had been gathered, and what needed to be done in the project.
Throughout the information gathering, I was evaluating what direction this project should be taken in. From talking to the operators at the facility, giving them another dashboard wasn't a great idea, since there are already a couple that are actively being used to monitor sensor data. After discussing it thoroughly with the project team, the rest of the robotics team, and my mentor at Cognite, we arrived at the conclusion that my time during the summer would be better spent updating and developing a feature that had been started on last year, but was sitting in stasis after the project it was designed for ended.
The way Cognite's robot control tool functions is that missions are defined, which contains a list of places that the robot needs to stop, and what sensor data needs to be collected from each of these "waypoints". Each time a robot executes a mission, it is called a run. The plan was for me to make a mini-app called mission results that appears alongside the other robot-specific tools offered by Cognite, to display the data collected during these runs in a way that is understandable for operators, managers, and other stakeholders that don't necessarily have a data science background. When explaining it to operators, it was useful to draw parallels to line-walks and routine inspections.
Since the project was changed to have a larger scope, and was no longer a client specific solution but rather a product, I held a workshop with people from the robotics team to define a set of users (rough personas shown clustered below). I then processed them after to reach a set of rich users that could be used to guide the design decisions. The main users were still the operators at the facility.
My next steps in the project were a combination of developing a flowchart and wireframing (first on paper, then digitally), and flipping between the two to help flesh each other out. This is where I spent a lot of time attempting to determine what were important features, and trying to stay aligned with the rest of the team on what should be prioritized.
Following the wireframing, the next step was to start designing in higher fidelity. When doing so, I like to focus on the functionality and site structure, making sure that things are built logically and consistently. I also discuss and spar with others as much as it makes sense to, since getting a fresh set of eyes on a design can be immensely helpful.
Having planned a thorough user test with one of the operators, what you see below is the prototype that was designed for the user test. The user test was very structured and I attempted to be methodical in my approach, collecting information about what is perceived, expectations, and usability for each screen or function I was testing.
Using the results of the user test, the rest of the summer was spent on fleshing out and polishing up the prototype, as well as creating and preparing for various presentations. Below you can see how the sitemap ended up, as well as the homescreen of mission results. The prototype can be shown upon request.
