Simulating “Mars on Earth” A Preliminary Report from FMARS Phase 2

Bill Clancey
Institute for Human and Machine Cognition, UWF, Pensacola
NASA/Ames Research Center, Mountain View, CA
July 22, 2001

By now, everyone who’s heard of the Haughton-Mars Project knows that we are here on Devon Island to learn how people will live and work on Mars. But how do we learn about Mars operations from what happens in the Arctic? We must document our experience-the traverses, life in the hab, instrument deployment, communications, and so on. Then we must analyze and formally model what happens. In short, while most scientists are studying the crater, other scientists must be studying the expedition itself. That’s what I do. I study field science, both as it naturally occurs at Haughton (unconstrained by a “Mars sim”) and as a constrained experiment using the Flashline Mars Arctic Research Station.

Over the past week, I lived and worked in the hab as part of the Phase 2 crew of six. Besides participating in all activities, I took many photographs and time lapse video. The result of my work will be a computer simulation of how we lived and worked in the hab. It won’t be a model of particular people or even my own phase per se, but a pastiche that demonstrates (a proof of concept) that we have appropriate tools for simulating the layout of the hab and daily routines followed by the group and individual scientists. Activities-how people spend their time-are the focus of my observations for building such a simulation model.

The FMARS Simulation

The FMARS simulation will be constructed using a tool called Brahms, which we are developing at NASA/Ames Research Center. The components of a Brahms model are fairly easy to understand:

• People (Agents & Groups)
• Geography (the building and its layout, furniture, etc.)
• Objects (e.g., generators, suits)
• Activities (e.g., reading email, EVA prep, watching movie, downloading & sharing digital photos)

Why do we want to build such a model? The number of applications may be surprising:

• Hab design
• Automation design & testing: Software Agents
• Formalizing analog experimental protocols
• Crew Scheduling and Communication-Coordination Planning
• Training astronauts
• Education – public relations
• Research on work and behavior modeling

Our immediate interest is to develop Brahms well enough so the various applications can be explored in further research projects. For example, through NASA funding we will integrate the FMARS simulation with an existing simulation of an air recycling system and an artificial intelligence monitoring and control system. The FMARS simulation will place loads on the recycling system, providing a contextual model of hab operations for testing the AI software. Furthermore, the (simulated) crew will interact with the AI software, for example, getting information about resource capacity (e.g., oxygen reserves) needed for planning daily work. In later work, we would like to develop computer programs that use a Brahms model to understand what the crew is doing, so the programs can provide appropriate support.

Methods

How do we build the FMARS model? There are two primary methods:
Participant observation (learning by being a member of the crew) and photographic documentation (including time lapse). During my week in the hab, I took regular notes about who did what, where, when, and why. Each day I added to this, refining with details, and finally developing hypotheses about why activities unfold in the manner I observed. In short, I need a theory of “what happens next.” What determines the next behavior of individuals and the group?

To organize my observations, I created a table in a document, with columns for the name of the activity, the location where it occurred, the time, who participated, and comments. For regular activities, such as EVAs and meetings, I used the table to record when the activity began and ended. By the fourth or fifth day I was able to sort the table more or less chronologically for a typical day and segment it into broader categories (e.g., breakfast, briefing/planning, EVA). Towards the end of the week, I began to refine some activities into subcategories (e.g., reasons for working at a laptop). Finally, after I left the hab, I realized the significance of activities and modes of behaving that I had not thought to write down earlier (e.g., listening to music while working at the computer).

My other notes were organized in an outline, organized by topics that emerged during my stay:

• Steps in an EVA
• Phase 2 Incidents (e.g., electric generator problems)
• Why staterooms are not used during the day
• Problematic spaces (e.g., the workstation area is cramped)
• Non-issues (e.g., staterooms are quiet)
• What can we learn from the sim as it is configured?
• Examples of how to analyze incidents to learn from the sim
• Ways to constrain hab activities for analog experiments

Also, at various times I wrote down where everyone was in the hab and what they were doing. This provides a snapshot of life in the hab. In retrospect, I should have done this on a regular basis (e.g., once an hour), for it would be a good way of verifying the simulation model. I had intended to follow someone every day, to note their behaviors in some detail, but as a participant in the hab, where group activities dominated (mostly organized around EVAs), this proved impractical. Finally, after I began to understand why activities occurred when they did, I realized I needed statistical information about events (e.g., how often and when we received radio calls from base camp). It requires more than a week to realize all that one might study (especially if psychosocial factors are included). Plus I believe that several weeks would be necessary to realize what categories are relevant; I am uncertain whether a crew member would ever have sufficient time to make and record all these observations.

Results

What are the results of my observations? I now have a table with about fifty activities, grouped according to broad “times of the day.” Here is an initial description of these broad periods during a day in the life of FMARS 2001 Phase 2:

• 0700-0900 Breakfast
• 0900-1030 Briefing/Planning
• 1030-1500 EVA
• 1500-1530 Eat and Clean up
• 1530-1700 Briefing and planning
• 1700-2000 Computers (email, photo download, software testing); data analysis in lab (CC)
• 2000-2100 dinner and cleanup
• 2100-2400 movies, refreshments (especially chocolate)
• after midnight: sleep, reading and writing

This outline is a broad abstraction, an average of seven days, not a schedule we followed. Nevertheless, the patterns can be striking. For example, on three sequential days the EVA crew stepped into the airlock at 1105, 1106, and 1108. No procedure required that we do this, it was just an emergent product of our intentions, the constraints of getting into suits and fixing radios, and our other habits (such as when we awoke, how long it takes to eat, and time to arrange personal gear). Absolute times will vary each day, but relative times, such as when a debriefing occurs after an EVA, are more regular (in this case, about 30 minutes). This chaining of group activities is a key part of the order of the day (which might be explained as part of individual, psychological processes).

What I have said so far should make clear why it’s not reasonable to expect a “human factors” report from the hab every day, providing research results. Unlike the biologists and geologists, I do not collect isolated samples in plastic bags. My daily observations are mostly too mundane to mention (as the pattern itself hardly seems surprising). Also, it takes four to five days until apparent habits are established, and then a few more days before details can be filled in (e.g., what are people doing for so many hours at their computers?).

Time Lapse Video Example of EVA Planning

An example analysis of a time lapse video reveals how I do my work and what can be learned. Based on an experiment in the initial year 2000 occupation of the hab, I placed my Hi-8 video camera in the far corner, in front of the right-most stateroom near the SE portal. I captured quarter-frame images (320 x 240 pixels) direct to disk every 3 seconds using a PC Card and video-editing program. Experience in analyzing such time lapses since HMP-1999 showed this frequency to be useful and sufficient. I captured two entire days in this way. In retrospect, I might have left the time lapse running every day. Full analysis is tedious, but the time lapse is also useful for capturing broadly the behavior of the group during critical periods, as the following example illustrates.

Before the EVA of July 15, the group discussed where to go. I was not part of the EVA crew, so I sat to the side writing, but also observing. When the group all gathered around Zubrin’s laptop, I began paying more attention and took photographs of the ensuing action. Similarly, the Discovery photographer picked up his video camera and began filming the action. The incident was immediately interesting because it illustrated a group planning activity, using multiple representations, coordinated with views through the portal.

Fortunately, we have a time lapse recording of this activity, so we can see all movements, who is initiating changes, and when the changes occur (within three seconds). The group moves like a flock of birds during a 12 minute period, gathering at the laptop Landsat image (Figure 1), NW Portal (Figure 2), a projected map of crater, and an air photo on table. Everyone participates. The commander tends to move in broad sweeps from one end of the floor to the other, to be contrasted with pivoting around the central area of the floor or tagging along. One person appears to be interacting with the commander most closely in these movements, suggesting a joint decision-making process. Another crew member lags behind with his coffee, but always joins the group to share their view.

Figure 1. Hab crew planning a traverse, gathered together to view a Landsat image of the von Braun Planitia on the commander’s laptop computer.

Figure 2. Hab crew planning a traverse, gathered together at the NW portal to view the von Braun Planitia directly.

Strikingly, the activity is clearly over when the commander, the crew member who was speaking with him throughout, and two other crewmembers stand to form a square and laugh. It looks like closure (Figure 3). The group then obviously disperses to prepare for the EVA.

Figure 3. Four members of the six-person crew, at the end of the decision-making process, now spaced around the room.

This example illustrates the value of having a time lapse going at all times for the sake of capturing such group activities. The photographs also illustrate that the conventional manner of documenting such activity (notice the Discovery Channel cameraman to the right of the group in Figure 1) fails to capture the overall pattern of how people are gathering and moving as a group. The time lapse shows very well how one or two people reorganize the activity by calling attention to different representations (the photos and maps) and the views out the window. Thus, the use of representations (including of course people’s utterances) is strongly contributing to individual attention, such that we can talk about a group activity. Finally, the time lapse provides a means of quantifying the duration and phases of the decision-making activity. The end of the activity is particularly well marked (Figure 3).

Layout of the Hab

An important part of the Brahms simulation of FMARS is a virtual reality depiction of the facility. The data gathered includes extensive photographs of all objects and areas, close-up photographs for color and texture rendering, and a scale drawing of the hab (Figure 4). This drawing shows the layout at a particular time, with the precise arrangement of laptops and chairs. The workstation area is the most obvious area where design requires improvement. The built-in table is not deep enough (about 24 inches) and is too cramped for six laptops plus a large server display (which hogs the most attractive area below the portal and blocks the view). Strikingly, one or two people used the wardroom table for working, echoing the conventional rows of workstations one often finds in mission control centers. The floor area is obviously spacious; another table might be placed at right angles to the first (between VP and the ladder area).

Figure 4. Flashline Mars Arctic Research Station, upper deck arrangement at 16:15 July 13, 2001. One chair is on the lower deck in the EVA prep room; location of laptops, etc. is accurate within three inches.

Activity Drivers: What determines what people do next?

The most detailed aspect of the Brahms simulation is a description of each activity as a set of conditional steps or alternative methods. That is, the conditions-when an activity is performed-must be specified. Given the table of activities (outlined above), we see that group activities are the main driver of behaviors in the hab, fitting the chronology of the day: Breakfast, Briefing, EVA, Debriefing, Dinner, Movie. That is, during this phase in the hab, individual behavior is constrained most strongly by coordinated group interactions. Furthermore, the daily EVA is the central, pivotal activity of the day, with meetings, preparations, and even meals occurring around it. This implies that the backbone of the simulation will be behaviors individuals inherit (in the Brahms representation) from the “Hab Crew” group. Each behavior in Brahms is represented as a workframe, which is a situation-action rule. In general, the situation (conditional part) of the key workframes for Hab Crew activities will specify either the time of day (e.g., morning briefing) and previously completed activities (e.g., the post-EVA briefing).

Interruptions are secondary driver of behavior, including: Radio calls (from base camp) or satellite phone calls (usually pertaining to our communications systems), systems emergencies (toilet, comms), hab maintenance (refilling the water reservoir, refilling the generators), and media interviews (conducted in the lower deck). Frequency information for the radio and phone calls might be determined from the time lapse. I did not have the time (or presence of mind) to systematically gather information about the frequency and timing of when these activities occurred.

Individual activity , behaviors that are individually motivated and performed alone, fill the remainder of the day:

• science data processing (e.g., analyzing dosimeter data)
• report filing (both individual reports and the daily crew report prepared by the commander)
• email
• digital picture processing (backup, sorting, sharing),
• chores (cooking, emptying garbage, etc.)
• personal hygiene
• taking photos inside the hab
• recreational reading

In summary, the conditions on activities are the group’s practice, interruptions (reactive behavior), and individual practice. Individual activities may be periodic (e.g., checking email), based on time, or based on remembering something you planned to do in the hab. Group practice is mostly chronological, but is also scheduled as required (e.g., cleaning the suits), chained (briefing follows an EVA), and reactive (e.g., handling emergencies such as an electrical short in a backpack).

What can we learn from FMARS as it is configured?

The 2001 FMARS field season has been planned for six phases in July and August. As I write, not even half of this activity has yet occurred. A report on the value of the hab for research and planning for Mars missions will require considerable analysis of diverse experiences, with different commanders, and experimental protocols. However, based on my single week in the hab, I will present my initial hypotheses about apparent research opportunities. These of course do not in any way represent actual plans or commitments by the project team.

• Communication Protocols
  • Enforce and document communications between EVA crew and support team mission support (perhaps simulated by HMP Base Camp with time delay) e.g., help for identifying a second site for the geophone flute. Investigate especially exchange of contextual information and instructions for using equipment.
  • Contrast: Commander serves as communications hub with world’s experts/advisors vs. specialists in the crew individually communicate with their own peers and private contacts
  • Work collaboratively with experts on Earth for data collection planning, analysis, and interpretation.
  • Facilitate and study data sharing among the crew (e.g., exchanging files using compact flash cards vs. a shared network)
  • Develop an expedition memory -what a prior crew leaves behind for subsequent crews: Photographs, articles, history of activities, maps (should mission support provide or supplement this repository?)
  • Improve communications between the EVA crew and hab support using radio and video to provide a running commentary (e.g., our ‘stuck in the mud’ EVA during phase two showed the importance of VOX (voice-activated radio) and video back to the hab
  • Radio from base camp was sometimes intrusive; will the hab reach a point of operational stability where communications could be time-delayed?

• Scientific Data Analysis Requirements
  • Determine the adequacy of laboratory equipment in the hab for data collection and analysis (e.g., rock slicing, microscope with camera)
  • Integrate computers used for data gathering and analysis with the hab’s communications and computing system
  • Determine laboratory space required and how to prevent intrusion from other activities

• Living and Working Priorities in the Hab
  

  
  As a rough cut, one can order priorities for the crew’s attention, based on basic needs and their interactions:
  

  • Electricity
  • Toilet (electric)
  • Water (pumped from outside hab)
  • Food
  • Privacy
  • Dry and well-ventilated staterooms
  • Quiet areas for sleep and work
  • Computer/lab work area
  • Personal storage areas (e.g., for cameras to be ready at hand)
  • Dry and warm EVA clothing (suits)
  • Cleanliness (showers, hot water)
  • Entertainment (e.g., DVD movies)
    

  The next step in this analysis would include comparing this list to published research, such as Stuster’s summary of findings from past scientific expeditions. A prioritization of needs and understanding of interactions is useful for future designs of space habitats, which will likely involve many tradeoffs and compromises.

Example of analog analysis

As I mentioned, abstracting lessons from incidents is not easy. Often the literal events are irrelevant, but a broader moral lies in the taken-for-granted context in which the event occurs. An incident during crew planning for an EVA illustrates my point.

The literal events are obvious to the observer: The hab crew is standing around the wardroom table, discussing how to set a GPS device for the planned von Braun Planitia EVA. Which GPS “system” should they use? The discussion concerned the nature of a GPS measuring system, and revealed that there were two alternatives available on their GPS devices (WGS84 [degree latitude and longitude] and UTM [metric distance]). One crewmember claimed that UTM was becoming standard; the people going on the EVA were more familiar with measurements in degree/minutes. They didn’t know how to use the GPS device to use or read UTM measurements.

The lesson to be drawn is not that astronauts going to Mars should be trained on how to use their equipment or that standards should be adopted before the mission begins. Everyone already knows this. Surely a funded, actual mission would have prepared the crew better.

Rather, the less salient and important issue is that route planning was occurring just before the EVA, not the evening before as I observed in HMP-1998 or as deliberately scheduled during HMP-1999 to coordinate with a mission support team in Houston. If mission support personnel were involved in choosing routes, then the crew wouldn’t be allowed to wait until just before departure to plan the EVA, this would have to be done the previous evening. So why did this group develop a different practice? My observation is that we were too tired from the day’s activities and too busy reporting what we had done to think about the next day. Thus, we may have a real issue here, which perhaps we thought we had understood in 1999. (It must be added, that the 1999 after-dinner communication with Houston focused more on reporting than planning.)

The example illustrates how a simple incident had to be interpreted in the context of previous field seasons, with background knowledge about past NASA practices and expectations. From this we see that FMARS provides a research opportunity for communications research, which can be exploited by enlisting more collaboration and establishing more formal protocols for hab activities (e.g., working more rigorously according to a schedule that is coordinated with Earth operations). Do we have the funds and committed external organizations (e.g., NASA, the Mars Society) to provide a realistic mission support role? Or should communications research using FMARS be focussed more on interactions between the EVA crew and the hab?

The GPS example also illustrates that superficial reports about FMARS operations are unlikely to give the viewer an understanding of what we are actually learning here or could learn from future analog experiments. Such analysis is especially the province of the participant observer and modeler who studies the scientists and operations in the analog setting, applying the methods and perspectives like what I have presented in this preliminary field report.

For more information about Brahms see

http://www.agentisolutions.com and papers at
http://WJClancey.home.att.net.