Srinath Sridhar Joins Brown CS As Assistant Professor
- Posted by Jesse Polhemus
- on July 21, 2020
Click the link that follows for more news about our historic CS With Impact expansion.
This fall, Srinath Sridhar joins Brown University’s Department of Computer Science as assistant professor. He’s the newest hire in the multi-year CS With Impact campaign, the largest expansion in Brown CS history. Currently a postdoctoral researcher at Stanford University, he credits his father, a self-taught user with contagious enthusiasm, for his first exposure to computing. But in some ways, that experience may have been less important than Srinath’s teenaged tinkering with cameras, which set the stage for his future research.
“I was compelled by the idea,” he explains, “of exploring cameras as mechanical devices.” A sense of wonder continues to be a motivating factor: Srinath has been photographing eclipses since 2003, and even traveled to Oregon to see one. “They’re celestial events I can’t postpone, something that’s much bigger than me, and that I’m not in control of. I find that fascinating.” And his early interest in the inner workings of things is still present. One recent personal project offered insight into how plants grow by taking time-lapse photos over many days, watching stems lengthen and leaves uncurl.
“I’ve always been interested in imaging,” Srinath says. “For me, deconstructing images is how I start to understand the world of humans and objects. They’re where my research begins.”
His primary interests, he explains, lie in 3D computer vision and machine learning, including overlapping topics in robotics, computer graphics, and human–computer interaction. “I develop 3D spatio-temporal machine learning methods that provide human-centric, object-centric, and interaction-centric understanding of our world from videos and images.”
Initially hoping to study computer science after high school, he ended up in geoinformatics, a branch of civil engineering concerned with gathering, processing, and delivering geographic information. Assignments in photogrammetry (extracting 3D information from aerial images of cities and other locales) were a powerful glimpse of what computer vision could offer, and by the end of his undergraduate days in Chennai, Srinath had decided that it was a worthy research topic. His next stop was the University of Michigan, Ann Arbor, for a Master’s degree, followed by a doctorate at the Max Planck Institute for Informatics in Germany.
“I was interested in 3D computer vision first,” he says, “but my other interests, like machine learning, came fairly soon afterward.” Among other things, the varied experiences showed Srinath the importance of mentorship: “I was lucky enough to be working in augmented reality before it was considered cool, and in ML before the deep learning craze hit. Part of the reason why that happened was because I had the right kind of mentors to help me figure out what would be interesting to work on.”
That kind of foresight, Srinath explains, is one of the reasons why he’s always felt at home in academia, and it’s something he’s eager to find at Brown. “I want to be around people who are researching long-term problems,” he says, “and the opportunity to teach a new generation of students who have radical ideas that I didn’t think of is wonderful.”
And the world that this new generation will inhabit, what does Srinath hope it will look like? It’s one in which robots would interact with their environment, and humans with robots, in the natural, unthinking way that we cook a meal or clean our house. We’ve gotten to the point, he says, where it’s easy for computers to beat us at chess or Go, but the challenge is human activities that we don’t associate with intelligence, like loading a dishwasher or folding laundry.
“One of my drivers,” Srinath tells us, “is to learn from humans in order to help robots perform these everyday tasks, and that means digitizing and examining human physical interactions from visual data–images and videos–so we can build smarter robots, and more immersive virtual/augmented realities. We can’t have them do what we do until we understand how we do it.” It’s a sentiment that recalls his younger self: wanting robots to function like people, Srinath’s analyzing human motion in the same way that he once did the innards of a camera.
“Grasping and manipulation appear so simple,” he says, “but we don’t completely understand how they work. Our hands played a massively important role in the evolution of human intelligence, but there are no robotic hands with human-level dexterity yet.” Over the next five years, Srinath hopes to expand his work on understanding human and object motion to encompass increasingly complex interactions. He’s also looking forward to multidisciplinary work with colleagues in cognitive psychology and other areas, seeing their expertise as vital for the next steps in human-robot interaction.
“My research is often inspired by applications,” he says. “We have pretty good robots and AR/VR headsets, but there are still major computer vision and machine learning challenges left to solve before these devices can interact like humans.” We will eventually have self-driving cars, Srinath explains, as well as realistic AR/VR and robots helping the disabled, but he’s not sure how long it might take for robots to have enough humanlike capability to free their owners from daily chores and let us do something more creative with our lives.
But he likes a challenge. “It might be decades away,” he says. “but I’d like to shorten that!”
For more information, click the link that follows to conctact Brown CS Communication Outreach Specialist Jesse C. Polhemus.