Robotics sculptor and machine learning enthusiast Timothy Noble (MArch ’23) probes power of tools, and tools of power

“Finding a satisfying combination of the technical and the social was elusive during my undergraduate years. I swung from Engineering major to History major and eventually wrote an undergraduate thesis about sustainable renovations to academic libraries amidst the dawn of the Internet and mass digitization of knowledge,” says Noble, who earned a BS in Sustainability, Environment and Information Systems from Antioch College.

“Honestly, sometimes I wonder how I avoided formally studying architecture for so long. But looking back, I see consistent through-lines that you could loosely group into an overall concern with spaces, their construction and meaning,” adds Noble, who grew up in the academic milieu of Big 10 college town Iowa City, IA.

"Architecture combines the social and the technical. It insists that people are generalists and can pick up different disciplines, to be conversant in them and include these perspectives in their work," he continues. "That’s what appeals to me maybe more than anything."

His first indication that architecture might be his intellectual home came in 2006, not long after he started his Intermedia and Digital Art (IMDA) Master of Fine Arts at the University of Maryland, Baltimore County. Through an email listserv on art and technology, he learned about an upcoming symposium at the New York Architecture League called Situated Technologies. It was co-organized by Mark Shepard, UB professor of architecture and media study, and Omar Khan, former UB chair and professor of architecture.

“Without exaggeration, I can say that it was life-changing. Artists, media-theorists, activists, and, apparently, even architects, were all wrestling with the explosion in computing, the proliferation of cheap sensors, and the diffusion of the Internet through Wi-Fi,” says Noble, adding that the symposium took place just before the introduction of the smartphone yet anticipated many of its societal effects and challenges.

The timing was just slightly off, however; Noble still had unfinished business as an artist. He would spend the better part of the next decade in an intensive making phase, brushing up against the field of architecture with inquiries in machine learning, robotics and digital media. Through sculptural assemblies of 3D-printed parts and open-source hacks, he explored the social impact of these very technologies, like the deskilling of labor and reinforcement of social class. In that time, he installed more than 20 solo and collaborative exhibitions and took on multiple teaching appointments from Baltimore to Bogotá.

“These works were an extension of my enduring interest in investigating process and the critical use of tools of production. I like to say my research exists between questions about the power of tools, and the tools of power. It’s also fun to make machines on the edge of functional,” says Noble, who conducts design and fabrication services for a range of clients through his ACME Spectacle LLC creative practice.

Among the highlights of his artistic career was a 2014 installation at the Burchfield Penney Art Center – “The Semi-Automatic Chalkboard” – a partial restaging of his MFA thesis that featured a drawing board using servo motors and open-source software to automate the production of Charles Burchfield’s sketches for his famous “Grain Elevators” painting.

“It’s about labor – an attempt to make visible the work behind an aesthetic artifact,” he wrote in his artist’s statement for the exhibition. “Today, garages and makerspaces accumulate technologies of remarkable precision and capability. But where does the human fit in this brave new future of automation? How do the social and the handmade persist within this increasingly coded landscape? Most importantly…what sort of literacies must we cultivate to maintain some agency in a world overrun at top and bottom by machines?”

In 2018, he completed one of his most significant artistic endeavors: “Of Lunatics, or Those Lacking Sanity” – a multi-media installation at Madrid’s Museo Reina Sofía exploring post-colonial legacies with the 2018 Goethe Medal awardee Mapa Teatro: Laboratorio de Artistas, an experimental performance art group. For the exhibition, Noble built three machines resembling artisanal gold mining carts, one that scaled the center void of a 100-foot stairwell and two that rolled dramatically down ramped sections of rails. The three were tied together with a system of 17 networked computers that included live "surveillance" video of the audience and multi-channel video documenting the mining community of Marmato, in Colombia's Caldas department. The project (installed in a space that originally functioned as a hospital for the mentally ill and was built in the late 1700s with proceeds from Spanish gold mines) served as a metaphor for the spiritually sickening effects of greed and exploitation.

“This was a sort of ethno-fiction portraying Spanish mining engineers with gold fever who came back from Latin America and had to be placed in the basement of the former hospital, where patients with mental illness resided,” Noble explains.

Still, Noble says, something was missing in his creative pursuits. “I am a maker, but my creative process was locked into iterations of physical making and revision. I realized I wanted to communicate my ideas through drawing, and that architecture was the best container for my skill set.”

UB’s research-intensive and interdisciplinary leaning three-year MArch, designed for students without an undergraduate degree in architecture, turned out to be an ideal fit. “Yes, my interests are in building and craft and how things come together, but UB had a program that had already gone beyond a narrow focus on buildings.”

“I wasn’t disappointed,” Noble says. “Students in the MArch program come from different places and backgrounds, and they bring that perspective with them to the program. The studio experience was different than anything I had ever experienced,” he said, referring to UB’s intensively collaborative studio culture. “Those core studios built my design skills. The architectural history courses were fantastic, and there isn’t one faculty member I didn’t learn something valuable from.”

Inherently inquisitive, Noble leaned into UB’s research environment, joining the School of Architecture and Planning’s Center for Architecture and Situated Technologies’ research team and pursuing the MArch thesis track.

Again, the 2006 Situated Technologies symposium served as inspiration, says Noble, recalling an off-hand comment made by a panelist, that Google Earth and its trove of satellite data could offer new and interesting ways of seeing the real world. “That comment was the spark that led to my MArch thesis.”

That, and the three years Noble spent in Bogotá – Colombia’s largest city and capital – where he served as a visiting lecturer and then as an instructor in digital media and computational design in the Facultad de Artes, Pontificia Universidad Javeriana.

His final MArch thesis, “City of Bars: Towards a Reading of Bogotá’s Ornamental Security Metalwork,” developed under the mentorship of Mark Shepard and UB faculty member and alumnus Albert Chao (MArch '11), catalogs and maps patterns in Bogotá’s ubiquitous and ornamental window bars and security grills using Google Maps and Google Street View (GSV) as a machine learning tool for object recognition and classification.

“Bogotá is intensive in a way that North American cities are not. I came to recognize a shared history of urban crime and violence whose legacy shapes the day-to-day practices of its residents,” Noble writes in his thesis narrative of his encounters with the city. “Gradually, I absorbed a certain architectural truth about Bogotá's buildings and its homes in particular: that all windows and doors on at least the first and second stories were visibly secured.”

Pervasive computing became the perfect tool to investigate this ubiquitous urban feature. The research endeavor required Noble to work nimbly across scales, tunneling into pixel data for an almost microscopic view of the city’s urban fabric, and zooming out to the sociocultural context of Colombia’s troubled and traumatic history in his search for patterns.

“Colombia had a very difficult 20th century filled with political violence, poverty-induced crime and the rapid urbanization of its major cities driven by the large numbers of citizens displaced from the countryside by armed conflict,” Noble continues in his thesis. “This friction and cultural diversity is reflected in the variety of metal grill patterns through which Bogotá[‘s] residents look out upon their city.”

Using a bespoke and hacked-together set of digital tools, he identified several hundred distinct ornamental patterns and developed a classification system to visualize stylistic groupings and tendencies on a neighborhood scale.

Noble’s methodology uses Google APIs (Application Programming Interfaces) to access GSV’s massive stores of location and image data and teach the software to identify patterns and make predictions.

“Google is essentially spraying three- to four-story buildings with lasers to capture point clouds - dimensionally accurate measurements - along with their photography. But since only the photographic imagery is publicly accessible, that’s what I’ve been working with. During my MFA work, almost 20 years ago, I was exposed to early computer vision tools like OpenCV. These were limited to tracking blobs of similarly colored pixels as they moved within a video frame, but they couldn’t distinguish between a cat and a dog. Newer tools like the family of machine learning models called YOLO (short for You Only Look Once) can learn to assess very specific characteristics, like the shaded borders that mark the edges of a window or door. Using a generically-trained base model, we can then retrain it with examples of specific objects and features that we want it to recognize." 

“For my purposes, I had to come up with examples of what I wanted the model to see by collecting façade imagery that teach it where windows are. After hundreds and maybe thousands of images, it begins to outline the windows and doors on its own. When we ask additional questions – e.g., to classify those windows and their bars – we feed it more images. The more results that are tagged, the more accurate the machine becomes.”

Noble says his research was undoubtedly influenced by the pandemic. “I had originally hoped to add depth and texture to the mapping and cataloging of these window bar designs through an ethnographic approach involving interviews with craftspeople and their clients in Bogotá, but travel was impossible. On the positive side, this deepened my interrogation of the street-view dataset.  It also pushed me toward a more disciplined review of the literature on Colombia’s complicated history of insecurity and instability.”

Sensing he had only scratched the surface of his research question, Noble has continued his thesis research, digging into the historic and sociopolitical forces behind the metal bars and their patterns across Bogotá’.

“I had come to read them as a sort of scar tissue, and it was apparent that I would have to shift registers and to grapple with the historical causes of insecurity and instability that the bars were symptomatic of.  Maps and Google Street View would only take me so far. Over the past three years, I have been visiting archives in both Colombia and the U.S. and working with documents, grasping at the causes and effects of the Bogotazo, a series of riots in 1948 that deeply and violently altered the trajectory of Colombia.”

Noble pays tribute to the faculty members and interdisciplinary culture at UB that have shaped the trajectory of his research: a precedent study by John Ringland, UB professor of mathematics, and Martha Bohm, UB architecture associate professor, on mapping roadside crops using GSV, which provided Noble with his baseline machine learning code; a brief professional appointment with Urban Planning Clinical Associate Professor Kerry Traynor and her preservation planning firm, which taught him techniques in public document research; and a course in GIS he took during his MArch with JiYoung Park, associate professor of urban planning, which informed his zoomed-out analysis of Bogotá’s barred condition.

“In a sense, work that began with some simple questions about the ubiquity and ornament of these quotidian window bars and whether it was possible to catalog their variety and distribution on a city scale, continues today with shifting methodologies that embrace the detective-like orientation of a historical researcher. At some point, I realized I was working on a book.”

Indeed, as much inclined to probe social context as he is to tinker with technology, Noble brings a critical lens on the digital-analog divide across his work as a scholar, practitioner, and educator.

"There is something interesting about the limiting factor of the term 'machine learning.' It can do a thing. That’s it," he says. "Apart from the breathless hype around AI, it puts the technology in its place."

As an adjunct faculty member, Noble is part of a nine-person teaching team supporting this year’s class of more than 110 first-year architecture students as they consider these same questions in their explorations of space and thinking through making.

For instance, students in the Architectural Alchemy first-year studio recently wrestled with a design exercise involving repeated 17-degree rotations of their assembly, a painstaking task that could easily be automated through a software like Grasshopper.

“We kind of need them to go through the difficult, analog version of the tutorial first,” said Noble, noting the essential value of architectural sketching and drawing and analog fabrication tools like the table saw or drill press.

“There is a real value to what we make by hand,” says Noble, who worked as a bike mechanic, carpenter and home builder in his 20s and once built a working 1970s Volkswagen out of parts. “I see students in the shop making simple cuts and holes with a CNC router. ‘Use the drill press,’ I want to say. But it’s also sometimes nice to be able to press a button and skip the tedious work. Being critical about when you use which tools is more the question.”

Noble, for one, found himself in need of a change from the technical labor of his thesis after completing his MArch in 2023. But today, he is approaching his inquiry in machine learning with fresh eyes. He is in conversations with faculty in both architecture and urban planning to apply his methodology to new areas of research – focusing Google Street View on different façade and streetscape features, for instance, or discovering hidden historic works of architecture. He is also carrying on a collaboration on the second installation of Mark Shepard's Hedgework project, a biodiverse hedgerow and urban landscape of native plants in New York City that interacts with multi-species visitors through a generative electroacoustic soundscape. Noble's part has been to develop small machine learning cameras that can identify (by species) birds and pollinators visiting the hedgerow.

Noble is sometimes bowled over by the fact that he now works alongside and in collaboration with the same faculty that dazzled his imagination and pushed his limits as a student.  

“I’m at an R1 research university within a School that operates at a manageable, accessible scale. I feel like I've found and been accepted into an intellectual home where I can work, collaborate and keep learning. This is a place where you can make things happen.”

Looking ahead into that wide open space, Noble says he will continue to meander in line with the same unquenchable curiosity that got him here.

“There’s probably a lesson in my work, in adapting to follow your curiosity where it takes you. If that curiosity is strong enough and your questions about the way the world works are sufficiently compelling, learning strange disciplines and new techniques are relatively minor obstacles. It becomes an exercise in adding new tools to your toolbox, which is always empowering.”