As part of a new project from PhD student Andrew Xu and computer science faculty Dr. Jacob Biehl and Dr. Adam Lee, cameras are popping up throughout the 5th and 6th floors of Sennott Square to investigate how public spaces are used.
Xu explained that the monitoring is done using USB webcams that are set up around high-traffic areas of the building, such as the elevator lobby on the 6th floor. The cameras are then connected to small, powerful systems called Nvidia Jetsons that use computer vision to extract data from the webcam about space usage. Though the data collected is visual, the data output leaves figures entirely unrecognizable.
“[This project] runs against another fundamental challenge of any type of technology in space, which is around privacy and security. The techniques that he (Xu) is using are computer vision based,” Biehl said. “All of the processing happens on that device and the only thing that the device transmits is information on occupancy. In particular, you can’t tell faces, you can’t tell identities — you can’t tell much other than use.”
A sample of the sensor output captured by the project.
The project, which has been approved by Pitt’s Internal Review Board, ensures that signage appears in every area being monitored, allowing passerby to know that the area is part of an active research project.
Xu said he initially conceived of the project during the fall 2020 semester, when he first came to Pitt in the midst of large changes in the academic world due to the pandemic. He wanted to answer the question of if and how people would change the way they utilize space if they had access to the data on how others use that same space. However, Xu said the implications of their research extend beyond usage due to COVID-19 in an academic setting.
“Let's say you're a student and you know that this particular area is high traffic and you want a more quiet study area. If you see that information about the current room you're in, you’re going to be like ‘Okay, maybe I’ll find somewhere else,’” He said. “That's one potential application of our research if it works out.
According to Dr. Biehl, the motivations he and Xu have for monitoring space use are informed by COVID-19 in both the short term and long-term. Short-term uses of spatial monitoring can help inform members of the University about the spread of COVID-19 and the importance of social distancing, while the long-term impacts COVID-19 has on the way people use space can similarly be monitored using this research.
“It was during the pandemic [that] we were kind of confronted with this observation that the University was making all these rules about how space should be utilized in the context of COVID,” Biehl said.
The short-term need for monitoring space came from social distancing and other regulations imposed by the University due to the COVID-19 pandemic. Biehl said that one of the ideas behind the project was to understand how well certain policies were being honored. Using artificial intelligence to monitor these spaces can also serve to reduce the cost of a policy audit that would otherwise have to be done using human labor.
“You can’t say a policy doesn’t work if it’s not being followed,” Biehl said. “We thought; ‘are there technology solutions to this where we can automate it?’ That’s the value of almost all computation is that you can take processes that are really expensive either for a machine to do, or, particularly, a human to do and then allow a machine to do it much cheaper.”
The long-term implications of the project are informed by the way work is changing due to COVID-19 and shifting models of labor. Biehl and Xu predict that hybrid work will remain after the major effects of COVID-19 are a thing of the past and, therefore, spaces that were originally designed to be used by all of their intended occupants — including students, faculty, and staff — every day will need to be rethought in a world where many people are physically present in these spaces for less time.
“Because the world is changing to a much more dynamic model of work, spaces themselves are going to become much more dynamic and we won’t be able to manage that if we can’t measure it,” Biehl said.