5 mins read 20 Sep 2022

Space Archaeology Expanding Technological Horizons

A new research paper from the International Space Station Archaeological Project outlines how archaeologists have helped develop and use machine learning to identify astronauts in ISS photographs.

The more “traditional” tools of the archaeological trade. Credit: Flickr/Dr._Colleen_Morgan.

Archaeology is not typically a field that most people think of when discussing cutting-edge technology. But, in fact, archaeology has a history of adapting rapidly to emerging technologies. For instance, archaeologists these days often capture geospatial and visual data using phones, tablets, and even drones. You won’t catch a modern archaeologist with a paper map very often; maps are now in a little flat box in the palm of your hand. However, one of the most adaptive, and unique, fields of archaeology is space archaeology.

Space archaeology is the study of humans and their interactions with the outer space environment using archaeological methods. This field of study is relatively new to the archaeology scene, but it is already gaining momentum due to its unique methods and its potential for application in future space missions. A new research article from Chapman University in California and Flinders University in South Australia is demonstrating just this by using AI to identify astronauts in photographs taken aboard the International Space Station (ISS). 

This new space archaeology article comes from the International Space Station Archaeological Project (ISSAP). This project is delving deep into the culture surrounding the ISS, from looking at what happens on the ISS itself to what becomes of the items removed from the space station. One of the lead researchers on this project is Australia’s Dr Alice Gorman (Flinders University), who has also been featured on several times for her work in space archaeology. Gorman commented on just how much archaeology has changed since she began her career in the field decades ago, particularly now with this new research being conducted. 

“Because we can't go to the ISS for fieldwork, we've had to come up with new ways to get the information we want. 

It's pretty wild to think that I started my career with a trowel in hand and now have exchanged this for software - although I have to rely on the expertise of colleagues like Rao Hamza Ali, who's the real expert in this!”

Examples of the images used in the ISSAP study. These photos show the ESA astronaut Luca Parmitano. Credit: NASA Flickr/ Ali et al. 2022.

Rao Hamza Ali is the lead author of the new ISSAP paper, which explored and applied a deep-learning based computer vision pipeline to automate the identification of astronauts in photographs taken aboard the ISS. With the ISSAP, Ali specialised in machine learning, which is the development of methods wherein computers gather data and learn in order to perfect the performance of a set task as much as possible.

 In this latest paper, researchers used machine learning to perform the task of identifying astronauts in photographs in order to simplify the task - one which would take humans a very long time, considering there were tens of thousands of images to analyse. In the end, the machine learning program reached 78.69% accuracy in identifying astronauts in the images correctly. 

The identification of the astronauts was then able to be mapped to form several insights into the astronauts’ social interactions and their structures. The results of the study found that crew who had been on multiple expeditions interacted with the astronauts from their first expedition more than any other crew. It was also found that the interactions were high between crew of the same expedition and low between crew of two different expeditions.

“I could not have imagined the potential of machine learning for analysing human behaviour in space. It's what my collaborator Justin Walsh calls a 'creative re-imagining' of archaeological methods,” said Gorman. 

An artist’s rendition of an Australian moon base. Credit: Australian Space Agency.

So why go to all this trouble - why try to identify astronauts from a huge pool of photographs? It’s because the ISS is so remote and such an unique environment, and therefore traditional archaeological methods can’t be easily applied (digging with a trowel aboard the ISS would not be useful). However, remote methods such as examining photographs can provide insight into what life is like in space.

“Living in space is very challenging because of the isolation, intense work schedule, and the amount of planning needed for everyday tasks which are simple on Earth. Crew on the ISS are expected to be productive to justify the costs of maintaining the station, so feeling part of a team is a big part of this.

“There have been many psychological studies of small, isolated groups such as Antarctic research scientists and astronauts, but our premise was a bit different - we wanted to look at how the crew used the internal spaces, and how this affected social cohesion. 

“Using the NASA image archive to see how the crew interacted showed that social ties according to nationality or agency of origin were less important than those created by working together on the same expedition,” said Gorman 

Data like this can be potentially used to design future space missions, perhaps in this case to figure out how to better foster social connections between expedition crews to create a better sense of being a part of a team. One organisation, Brick Moon which was recently  co-founded by Gorman, plans to utilise perspectives from a range of disciplines (including space archaeology) to create orbital or planetary habitats that improve productivity, reduce costs, and support crew well-being. This means that archaeology could be used to transform space habitation as we know it. The field has come a long way from the humble trowel.