Archiving Robots: interdisciplinary collections & interdisciplinary teams
I was the Principal Investigator on a grant from the Sloan Foundation that explored the challenge of archiving robots. We had a great team and lots of inspiring conversations. Here are some highlights:
What is robot?
This was a question that we wrestled with early on.
For some it was the machinery. The device. The body. The design. The thing. The hardware. This could be considered the classic engineering mindset.
For others it was the code. The algorithm. The brain. The software. The action behind the movement. This was more of how computer scientists thought about the field.
And so, with the idea of building an archive, we toggled through that physical and digital world. We landed on the term “multimodal collections” which we abstractly defined as the interconnected ecosystem of tangible and intangible information, objects, digital artifacts, and narratives that comprise the scientific research process
This table is helpful to visualize the scope and scale of what we encountered:
Which version matters?
We also talked a lot about versions. For instance, version 1, version 5, or version 10. Which one should we focus on? Did we try to document them all? Was the first one the most significant? Or did the last one represent the overall culmination? Or was the middle one actually the most critical because it introduced a transformative new method that enabled substantial progress?
Also: did we want to restore robots? Or did we try preserving them within a certain historical condition?
Even more complicated: we discovered that parts were often cannibalized. A sensor from one robot would often be redeployed on a completely different project. There wasn’t a lot of documentation. It is a very spontaneous and iterative enterprise.
Process Over Products
A guiding mantra that emerged for us was “process over products.” The spirit was to capture context and stories not just parts and pieces. If you have a circuit board, some wires, and a set of wheels— they might not mean much without the backstory. We didn’t just want to fill a warehouse with stuff – instead, we aimed to understand robotics as a scientific endeavor. How did the discipline evolve? How did breakthroughs happen? What did collaboration look like? Why was version x important? How did teams overcome setbacks? How did they organize their work? What was the initial goal, mission, or aspiration – and how did that evolve alongside the growth of the technology? We were hyper curious!
The Robotics Research Ecosystem
As we attempted to better understand robotics as a culture, we recognized that it didn’t just occur in the lab, but that the work was situated within a larger ecosystem. And to tell the story properly we needed to be inclusive. Here is a quick sketch to illustrate some of the key roles:
Our Interdisciplinary Team
What I enjoyed most was the interdisciplinary team that emerged around our theme. It was an organic ad hoc community. We pulled in diverse expertise to help us explore and tackle different challenges. I loved digging into the variety of mental models and disciplinary toolkits.
For example, from geology we compared robotics to stratigraphy. The rock layers stacked upon each other over time was like the digital and physical layers of code and components we found in the labs.
We also talked chaîne opératoire (operating chain or sequence) which helped us analyze technical processes alongside social activities.
Based on my notes, our community of expertise looked something like this:
Archivists
Librarians
Historian
Inclusivity Consultant
Software Developers
UX Designer
Physical Conservators
Digital Conservator
Information Architect
Data Specialist
Digitization Specialist
Metadata Specialist
Anthropologist
Archeologist
Marketing & Advancement Professionals
Facilities & Business Office
Project Manager
We also interviewed students, staff, and researchers in a handful of robotics labs to observe their work and to learn directly from them as well.
Preserving the un-preservable
This was another constant theme in our conversations. Ultimately, we were dealing with ephemeral constructions. Yes, they were metal, plastic, wires, sensors – tangible things – but they were never intended to last beyond the life of a project or the life of a grant.
We looked at time-based media but also at Fine Arts: theater productions, artist showcases, and music performances. You could document and record those types of activities, but it wasn’t the same as experiencing them live. When you’re attending a play — the interaction between the music, lightening, mood, and setting — that all works together to support the actors and the story…. it’s not the same as reading the script or watching a video. Robotics feels that way too.
What to collect?
This is the BIG question. The answer: it depends! We outline this in the toolkit (link below) but it comes down to so many variables. Space. Time. Capacity. Funding. It’s really about setting an intention and then shaping the related collection development strategy. We offer guidelines for unpacking that question.
This was a lot of fun. Special thanks to everyone who helped and especially to my colleagues Katherine Barbera and Kathleen Donahoe who kept things moving forward and asking provocative questions.
More about the ongoing work: CMU Library Robot Project.
Here is the paper: Multimodal Archives: A Toolkit for Collecting Robotics and Other Complex Material in a Research Ecosystem
And here is our CNI talk: Robotics Project: Insights
