Process
The research process that the team followed consisted of five steps:

• Collect
• Review
• Model
• Summarize
• Present

First, we collected data by either conducting a contextual inquiry or a post observation interview.  We recorded the data with video, screen capture, audio recording, and written notes.  Second, the team reviewed the data with analyzing software internally developed for the project.  This software allowed us to watch and listen to video and audio recordings and easily attach notes to specific parts of the recordings.  The software streamlined the note-taking process and greatly facilitated the modeling of the data.  Third, the team modeled the data by creating one or more of the following models: 

• Artifact model
• Cultural model
• Mental model
• Physical model
• Sequence model
• Work flow model

We determined which models to create depending on the data collected.  For each set of data, we summarized the data by creating a list of key-findings.  We then presented the data to both our faculty and to our client, who was non-collocated at the time.

Approach
Before embarking on our Contextual Inquiries, we wanted to make sure we knew what we’d be getting out of them. We knew that few, if not none, of the robotic domains we’d be exploring would be exactly analogous to the MER domain. Some robots would be fully autonomous with no human-in-the-loop; some would be commanded in real time and by a collocated operator; some would serve a purpose other than collecting science data. In order to overcome these discrepancies, we developed a matrix to organize the possible domains to explore and the features they shared with the MER domain. This would allow us to focus our CIs and know what elements we were looking for ahead of time.We first researched robotic taxonomy in order to gain an understanding of what kinds of attributes we should be using to establish congruency with our domain. We also did some initial investigation into robotic research in the Pittsburgh area that we might have access to. The matrix, which we continued to refer to and update throughout the semester, is shown below.

After reviewing our initial matrix, we recognized a need to look outside the robotic domain to gain a more fundamental understanding of in-situ re-tasking in general. We brainstormed potential areas where we could gain access to this type of work, and decided on nurses in an operating room, security dispatchers, and a fast food restaurant manager. Stepping outside the robotic domain also prevented us from developing ‘robotic blinders’, which would have had the potential detrimental effect of restricting our design ideas to those that are established and standard in the robotics field.