Interview with Grit Ventures Academic Council Member, Dr. Justin Hart, Associate Professor of Computer Science at UT Austin
Written by Melanie Landesberg
Grit Ventures University Fellow, Yale College Undergraduate
Dr. Justin Hart, a member of the Grit Academic Council, became interested in robotics in a manner familiar to many: he started programming computers as soon as he learned to read and write. Eventually, this interest in computers extended to artificial intelligence and robotics. Now, as a professor of computer science at UT Austin and assistant director of Texas Robotics, he is actively working to bring the robot-filled society of his childhood dreams to reality.
Living LARG at UT Austin
At UT Austin, Hart is a professor with the Learning Agents Research Group, which focuses on studying human-robot interaction for home, workplace, and public space applications. He studies autonomous human-robot interaction that will eventually put robots in people’s homes, offices, and public spaces where they can be of service and work alongside people.
One interesting project he described modeled how robots may interact with people when passing each other in a hallway. Currently, robots follow the middle of a hallway and operate with little consideration for people in their surroundings. Hart’s research group is working to imbue robots with the ability to understand human body language and “gaze,” which they can use to interpret someone’s intended path and move accordingly. Robots themselves may also ultimately display “body language” understandable to people so that each party of the exchange (robotic and human) can participate in the maneuver. This is not a limited use case — if robots can interact seamlessly in a hallway, robots can be more easily placed into offices, college campuses, and more.
Scaling Texas Robotics
So how does UT Austin compare to some of the past universities he has attended? Hart completed his PhD at Yale University, working under Professor Brian Scassellati studying artificial intelligence and human-robot interaction. He also held postdoc positions at the University of British Columbia and UT Austin, and was offered the non-tenure track position that he now holds. In terms of the scope of robotics projects, none come close to UT Austin. Hart is involved with the Building-Wide Intelligence Project, an initiative to deploy robots on UT Austin’s campus that interact with students, provide directions, carry out deliveries and more. In describing the size and scale of robotics at UT Austin, Hart said, “I’m the assistant director of Texas Robotics, which currently has 15 or 16 different faculty members… and we are still growing. Right now, we’re able to take on much larger projects than have typically been taken on by universities before… we have this large institute that allows us to increase our level of collaboration between faculty. That’s really key.”
In addition to a great academic and research program for robotics, UT Austin also provides many resources to bring inventions from lab to market. Some notable startups to emerge from UT Austin include Diligent Robotics, Apptronik, Cogitai, which is now Sony AI North America, and Spark Cognition. These companies originated with professors, but that does not mean innovation is limited to the labs. Briggo, a startup developed at the business school, is a robotic coffee maker that was acquired by Coca-Cola and is already installed in several airports and convention centers.
Vision: Humans and Robots Mingling
At the end of our interview, I asked Justin what his vision of robotics would be in the next five, ten, and twenty years. In some ways, he said, it’s a difficult to apply an actual date to the rate of innovation. The better question would be to identify the research challenges to overcome in order to bring capable robots into homes, offices, and public spaces around the world. In the next five years, Hart said, “I think in the next 5 years we’ll start to see the first human-robot interaction software packages that solve an HRI problem. The one I’ve been focusing quite a lot on is the social navigation problem.” In ten years, we may see single purpose robots that operate in much less structured environments than they are allowed now. For example, this may include factory floor robots that are not strictly confined to one space and are allowed to mingle with humans and other robots on a factory floor. The advances in technology will allow them to operate more independently with fewer safety concerns. But for more advanced robots that will operate in the home? “The timeframe for robots in people’s homes or for more general tasks will be a twenty-year goal,” he said. “I hope it’s sooner and I’d like to contribute to it being sooner. We’ll start to see the emergence of service robots in the next 10 years, though.”
For more information about his research, visit his website at http://justinhart.net or https://www.cs.utexas.edu/people/faculty-researchers/justin-hart.