Design Iterations

Overview

Testing of our design ideas was a critical task for determining what kind of interfaces were most effective for helping people plan rover missions. Throughout the last 4 months of the project, a continuous test and development cycle as used. Initially starting with several kinds of low fidelity prototype and eventually working towards a fully interactive robust application.

Lo-Fi Prototypes

Paper Prototype #1

We performed a think-aloud study, using paper prototypes, with four users. Each test participant was given four tasks that came directly from the analysis we did from our playtesting and makestorming sessions’Äîtwo from the preplanning phase, and two from the gameplay phase. The tasks were:

  1. Create a simple plan
  2. Add to an existing plan
  3. Access a picture that was taken
  4. Annotate target values onto a map

In general, these studies proved the success of our interface concepts, as all the participants were able to complete their tasks. Here are some of the major findings:

Insights:

  • Context-menu based interaction intuitive after discovery
  • Context menus should have all options including default
  • Macros such as driving in facing direction should be supported
  • Rover icon afforded direct manipulation that wasn’Äôt intended

Paper Prototyping #2

We performed a second think-aloud study, using paper prototypes, with a total of 4 users. This prototype was an iteration from the previous paper prototype. Each test participant was given tasks, which are as follows:

  1. Capture an Image
  2. Modify the Path
  3. Access a picture that was taken

Insights:

  • Image wedge based interaction intuitive
  • Failure to understand that rover was not capable of absolute positioning
  • Failure to comprehend the lower-right picture viewer
  • Labels are Misleading

Cognitive Walkthrough with Non-expert Users

Our think-aloud studies were difficult to conduct due to the high amount of cognitive load required to wizard-of-oz a computer interface on paper. We still wanted to have user data to test our concepts, so we modified the cognitive walkthrough to include users to test the perceivability of interface elements. This way, we could validate concepts generated in our early iterations of design. (Although, making sure the user was in the right mindset was tedious, we got good data). In addition to identifying users and their skill set, we needed to recruit and educate them about their skill set. We came up with our storyboards, but we needed to separate each screen so that we can find out what the user is thinking at each step without biasing their thinking. Also, instead of showing just the current and after state, we needed an extra state to show the action that occurred in order to find out if the appropriate feedback from the action made sense (fourth analytical question).

  1. Capture an image at the corner of the target
  2. Add a segment to the path to a given point
  3. Modify Path by adding another image at a target
  4. Delete a segment of the path
  5. Uplink commands after downlink