Role
Carnegie Mellon University Human-Computer Interaction Major Capstone Project
Project Overview
The goal of the CASC project is to build automated tools for assisting in harvesting specialty crops. Specialty crops such as apples are crops that require more attention and care to reach their full potential.
We worked with a CMU Robotics research team that was developing an automated prime mover (APM) that could drive itself around an apple orchard while helping workers carry out their day to day tasks.
You can read our complete project report here
Responsibilities
I was the technical lead for our project team and responsible for coding our prototypes and working with our client developer to figure out implementation solutions. The whole team helped to conduct all phases of research from contextual inquiry, through final user testing, and also to synthesize findings into various models and design iterations.
The APM (Automated Prime Mover)
The original developer interface for the APM.
To start our project off, we used a number of research methods including competitive analysis, contextual inquiry and heuristic evaluation. We synthesized this research into consolidated models and extracted a number of initial design ideas.
Some key design criteria came from this such as:
- Workers are very often in direct sunlight, so displays should be high contrast
- Our users will often be wearing gloves so physical controls are preferable to touch screen controls. Controls should also be large to help to avoid errors as the APM moves over bumpy terrain
- Many operators of the APM may speak little or no English, so the interface should accommodate multiple languages or should use symbolic representations that are universal
After our initial round of research we underwent 3 rounds of design iteration and testing.
The first round of testing was with a low fidelity paper prototype to validate our initial ideas. The second round of testing was with a high fidelity prototype that was connected to a set of data to simulate the movement of the APM. The final round of testing took our next iteration to the field with participants sitting aboard the APM piloting it down a row of trees while completing tasks using our interface.
Low-fi paper prototype explorations for initial user testing
Hi-fi prototype iterations
Lo-fi exploration for vehicle controls focusing on autonomous task
After our initial round of testing, our clients asked us to add some additional controls for configuring how the APM would drive down rows of trees. The above shows some of our design explorations for adding these new controls
Hi-fi iterations on the control bar. The key changes were made to clarify the controls for how far from the center of the row of trees the APM should drive.
User testing on the APM
