The AI integration into the user interface had applications across multiple areas including voice control, attention direction to determine the best path for the EV, information tracking, warning notifications, and EV to PR data translation that records conversations with the AI assistant as well.

The voice assistant was cleverly designed as an icon that visually reacts in modes of listeing, thinking, and responding back in a pulse like feedback animation.

Attention Direction was designed to redirect the EV in case of a warning that impacted distance capacity for the vehicle such as low battery.

Predictive Telemetry is a tool that can be toggled to reveal the telemetry of the EV or PR when on an ongoing route. This is a helpful tool to be aware of battery life, oxygen level, and cabin pressure.

Communication History was created with the purpose of recording all conversations and using the AI assistant to quickly categorize and repeat/recall important topics that were discussed.

How do we decide which features of the AI assistant need to be visually present in the interface and which features stay productively hidden? We considered the various features that AI could optimize for travel and wanted them to visualize as distinguishable AI features.

The AI assistant had the potential to optimize PR and EV navigations. However, populating the map with multiple alternate paths could increase the mental load of the astronauts.

Features like the voice assistant would occupy too much space within the screen and cover the map. Additionally, there was no extra space on the task screen, which meant the communication history would have to appear on the telemetry screen.

To efficiently distinguish AI features from non-AI features, a sparkle-like icon appears to identify the AI features across the interfaces. Additionally a toggle was mode added instead of a constant overpopulation of predicitive information for each POI (point of interest).

Considering the enviroment in which the EV and PR would have access to the voice assistant (a nondisruptive environment which is likely available), the voice assistant would not serve as a productive tool if all conversations populated into the map screen. Thus the communication history could be recalled from audio by addressing the voice assistant meaning that notifications of the audio assistant's answers were no longer necessary.

The predictive telemetry was not as informative in earlier iterations due to the worry of space optimization so telemetry notifications that were predictive were reduced to colored dots. This was difficult for users to recognize as important or how it directly impacted the user.

It was later decided that predictive telemetry had to be explicitly informative and recognizable as "predictive" which was solved with a toggle and hovering bubble over significant POI's on a path.

The dedicated space of communication history on the telemetry screen was incredibly limited and difficult to format in a user-friendly way. This image displays the initial formatting which we realized through user testing was difficult to spot and understand the controls of.

It was later understood that having the search bar be more prominent in the design was important to navigating through the communication history. As well as having the tagging system be visibly clear.

Through this project, I discovered that product design typically does not follow one goal. It is a multifaceted collection of goals that the user must know to have available and accessible to understand.

In previous years, physical limitations challenged the user interfaces we created as astronauts with large gloves and VR interfaces faced an entirely different set of issues. Even in the lack of this challenges, the AI assistant proved to be equally challenging and interesting to incorporate into a dense interface.

Suprisingly, I learned an extensive amount about the interface I was working on for months through other users who were looking at the first time. They confirmed some of my worries but also brought upon issues that I would not have considered to be impactful to the user. For our small team, we interviewed around 10+ professors and friends which was incredibly helpful to our final design.

Although I expected a restricted amount of creativity within the interface, I was surprised with how much control I was able to implement into the design. While staying within the visual system, I found interesting design challenges such as finding visible spaces for the AI assistant to display!