Overview

Problem

To prepare for Artemis lunar EVAs, EVA planners in NASA Mission Control need to understand the temporal and geospatial factors and constraints that may affect the mission, and create plans accordingly. During the EVA, they should be able to quickly suggest modifications if necessary.

Solution

Automated Spacewalk Task and Route Advisor (ASTRA) streamlines the EVA planning process by providing automated, optimized paths as a starting point for path building. ASTRA gives users the ability to compare, refine, and summarize their path decisions. Through this process, EVA planners build confidence, crystallize their arguments, and prepare to advocate their decisions to their peers.

My role

Lead Product Designer, UX Researcher on a team of five

Scope

8-month capstone project with NASA Ames Research Center

Live View

During Extra-Vehicular Activities, EVA planners must monitor for any current or future deviations in the EVA from the original planned path. Live View enables planners to see where astronauts are, and monitor how they are progressing along their route.

Data Layers

In order to better visualize constraints and geospatial information, GIS data layers are integrated into the lunar map view. GIS data layers can also be leveraged to visualize constraints, as seen in the Metabolic Advisor.

Off-Nominal Notifications

If an EVA is subject to change, ASTRA will surface an off-nominal notification in the Live View map. Off-nominal notifications can indicate varying levels of severity. This visual, paired with existing comm loops, enhances an EVA planner’s understanding of plan changes.

Waypoint Selection

When generating a new path, EVA planners must first select the waypoints that they expect the astronauts to cover. When beginning from an off-nominal notification, the New Waypoint, as well as the original path’s waypoints are pre-selected.

‍

Optimization Functions

With waypoints selected, the EVA planner may choose an optimization function to create their path options. Path building algorithms informed by optimization functions have previously been utilized in rover missions, and can be expanded to human EVA operations.

‍

Optimized Route Generation

The system is able to auto-generate paths according to a specified optimization function and waypoints. This automation eliminates impossible (i.e., unsafe) routes, allowing EVA planners to focus their time and energy on choosing between the potential compatible futures.

‍

Integrated Data

Route information is organized in a table format, allowing EVA planners to quickly and easily compare route options’ statistics. Additionally, there are indicators to differentiate between values that are safe, approaching danger, or completely dangerous.

Path Selection

As a planner compares and narrows down, they can deselect paths in the table to hide them from the map. This enables EVA planners to make more nuanced comparisons as they finalize their decisions.

Refinement

The optimization functions that inform ASTRA’s automated path generation rely on physical constraints, or constraints that have clear safety & feasibility parameters. For more nebulous constraints, ASTRA enables planners to use their judgement to make informed modifications.

‍

Path modification leverages existing GIS design patterns, with editing tools EVA planners are already familiar with. While ASTRA does give the freedom for planners to edit paths, the system does make it clear if a proposed change will violate a constraint identified by the optimization function. For example, if a path modification pushes the expected CO2 saturation beyond an acceptable margin, ASTRA will make this visually evident to the user.

Visualized Plan Change

Once an EVA planner reaches a path modification decision, an overview page of the changes of the new path as it compares to the original path will appear.

Problem Space

NASA and other international space agencies are experienced in extravehicular activity (EVA, otherwise known as spacewalks) planning and execution in microgravity. However, in the entire history of space travel, only twelve humans have ever stepped on the surface of the moon, in total completing fourteen lunar EVAs.

‍

While Apollo EVAs explored various locations around the lunar equator, future Artemis missions seek to explore and establish a sustained presence on the lunar south pole. This presents additional EVA challenges, most notable of which will be extreme lighting conditions and drops in communication availability, both of which change over time.

‍

To prepare for Artemis lunar EVAs, planners need to understand the temporal and geospatial constraints that may affect the mission and create plan changes accordingly. During the EVA, they should be able to quickly suggest EVA plan modifications while ensuring the safety of the crew.

Literature Reviews

To help us get started, our clients at NASA Ames Research Center provided a variety of literature on EVAs. At this point, our familiarity with space was limited to its vast and uncharted nature. After strategically dividing and analyzing NASA's extensive collection of research papers, we were able to piece together the current state of the EVA landscape. From there, we expanded our research scope and dove headfirst into the problem.

Expert Interviews

One of the most exciting opportunities we had with this project was the ability to interview 20+ Astronauts, flight controllers, and NASA engineers. Through these interviews, we were able to gain first-hand insights into how decisions were being made during EVAs.

‍

Working in sprints, we synthesized our interview and literature review findings by creating affinity diagrams to derive insights.

‍

"I was frustrated because communication for common ground was not clear and you have to . . . negotiate and have this conversation about it." ‍
- NASA Research Engineer

Analogous Domain Research

Our team needed to bring fresh perspectives into this huge problem space, so we selected professionals in analogous domains who face dynamic time and location constraints. Each interview was tailored to draw parallel findings to real-time, high-stakes planning that EVA Planners experience.

To further immerse ourselves in the problem space, we visited the Moonshot Museum and met with the IRIS rover team in the CMU Robotics Institute.

Park Pretotype

Since there is no way for us to experience an actually lunar EVA, we simulated our own EVA planning and execution in a public park. Half of our team assumed the roles of Mission Control, stationed at our office, while my teammate and I took on the roles of astronauts who were transported to the "Hab" (habitat) located within the park. From the Hab, the astronauts embarked on a mission through the park by completing “tasks” outlined by an EVA timeline, navigating the terrain, and keeping an eye on their consumables. Meanwhile, Mission Control acted in a supporting role by updating the timeline, maintaining crew safety, and solving problems.

‍

We learned that:

1. Astronauts had a different understanding of the environment compared to our MCC, making it complicated to communicate accurate instructions.

2. In our MCC, every officers’ actions impacted everyone. It was important that we stayed on the same page throughout the mission.
‍
3. The original mission plan quickly became irrelevant due to unaccounted obstacles and inaccurate route predictions.

The paper map MCC used to guide the astronuats

Reframing the Problem

We discovered a tension between the culture of safety and the need for efficiency in Mission Control Center. Given that safety is the utmost priority, efficiency is often compromised to guarantee that everyone is confident in the safety and optimality of the decision.

‍

Consequently, we sought out to find a solution that refines the planning process while upholding the safety of the crew.

Our Focus

After examining the existing EVA information flow and operational requirements, we formulated questions that would serve as the foundation of our project:

‍

How might we increase individual confidence and decrease frustration during tactical planning?
‍

How might we prepare EVA planners to advocate for their decisions effectively and confidently?

Insights

EVA Planners and astronauts have different mental models of the EVA environment, resulting in a disconnect that affects tactical planning.

We found that astronauts and EVA Planners have contrasting understandings of the map and environment. While astronauts rely on their personal observations and sensory inputs to navigate their surroundings, EVA Planners rely on technical data to create understandings of the astronaut’s environment.

In addition to this difficulty, unexpected changes in the crew's location or movements can also arise due to changes in mission objectives, obstacles, or technical issues. In such situations, the EVA Planner must quickly adapt and adjust the plan to ensure the safety of the crew and the achievement of mission objectives.

‍

EVA planners rely on manual tools for calculations and scratch work, resulting in information silos and a lack of data governance.

‍

EVA Planners use a variety of different tools in their workflow, doing scratch work in tools like Excel or utilizing pen and paper. Others in MCC are unable to see the individual work that was done, which leads to work not being properly documented or shared.

‍

EVA Planners have difficulty justifying and advocating for their work, which results in communication redundancy across Mission Control.

‍

Based on our interviews with NASA experts, we found that the process of adapting plans is non-linear in nature. There are extensive discussions that take place, with ideas going back and forth, before a final plan is agreed upon and implemented. In this process, we learned that EVA Planners often have to justify their work to ensure the validity of their reasoning.

‍

Storyboards

To test the feasibility of our early ideas and concepts, we developed a set of storyboards which we shared with experts to gather their feedback, capturing both their enthusiasm and concern. The following are a few of the storyboards we crafted.

‍

User Flow

We identified the two main types of deviations - nominal (within the acceptable range) deviations and safety-related deviations - and mapped out user flows for the two use cases.

‍

Prototyping

We made annotated wireframes to understand the key screens necessary for our final prototype flow, and subsequently translated them into interactive, high-fidelity screens. Throughout this process, we ran usability tests and worked iteratively to refine the screens.

‍

Concept Validation & Testing

We tested our design with 9 NASA personnel and experts with experience in EVAs and related fields. We weighted each participant according to their level of direct experience working with EVAs.

Weight of 1: NASA personnel with no EVA experience

Weight of 2: People with relevant experience, such as in NASA EVA Simulations or GIS

Weight of 3: People with multiple experiences with NASA EVAs

Procedure

Participants were given a scenario that places them in the middle of an EVA, where an incident happened that required re-planning of the remainder of the EVA. With consideration of specific factors and safety rules, they then had to decide on a new route twice: once using the control design and once using ASTRA. Our ultimate goal was to see whether our solution decreases negative workload-related emotions and increases confidence in advocating their decision

ControL

To mimic the current manual tools and fragmented data that EVA planners currently rely on, our control was a spreadsheet with a series of data of an EVA divided up by several tabs.

EXPERIMENTAL

Our experimental design was the first iteration of ASTRA, where all the potential routes were on one map and associated data was laid out in a table.

Methodologies

Think Aloud
We asked participants to think aloud while they decided on a route to help us better understand their thoughts and feelings during the decision-making process. This allowed us to gather qualitative data on how they felt making a decision.

Confidence Rating
After choosing a route, participants were asked to rate their confidence in their decision on a scale of 1 (not confident at all) to 10 (completely confident).

NASA Task Load Index (TLX) Questionnaire
Participants were asked to complete the NASA Task Load Index (TLX) questionnaire to assess their subjective workload ratings for our tool as it compares to the control.

Findings

In our tests, we qualitatively and quantitatively measured confidence & workload metrics of ASTRA’s users in comparison to their experiences with the control. During observation, we noted fewer instances of hesitation, more confident body language, and more certain phrasing when justifying responses to the experiment’s EVA Officer. In our post-test interview , users expressed positive feedback when comparing ease of use and access to information in relation to the control test. Furthermore, users who had actual experience in EVA back rooms shared positive comparisons to existing workflows.

‍

To bolster these observations, we utilized TLX questionnaire and confidence rating metrics to help us quantify subjective experiences reliably and consistently. Here are our results:

NASA TLX Subjective Workload

In the context of the EVA workflow, we believe this 26% decrease in mental load will result in more mental energy spared towards optimal decision making. By making it more difficult to design a plan that violates clear constraints, ASTRA allows EVA planners to focus on decisions that require human judgment.

NASA TLX Manual >

Confidence Ratings

In our research, we found that difficulties expressing certainty in a modified plan were responsible for lost time and increased frustration in the chain of command. This 15% increase in self-confidence supports our belief that ASTRA’s guided path building process helps to crystallize a planner’s line of thinking, resulting in more confident advocacy of a modified plan.

"I hope I get to use a tool like that"
- NASA EVA Systems Engineer

‍