Embodied AI Research Takes Center Stage at San Diego Robotics Forum

Quan Vuong, a Ph.D. alumnus from Christensen’s research group, is one of the people seizing this moment. He cofounded Physical Intelligence, a startup that aims to develop a generative AI model that can control any robot and get it to perform any task. He was the keynote speaker for this year’s forum. He detailed how his company has collected thousands of hours of video of robots manipulating daily objects as training data. The company has chosen to make its models public and open source.

“The internet of robotics data does not exist. Collecting large scale, diverse robotics datasets and making sure they are useful is a real problem. We are going after both these challenges,” Vuong said.

Meanwhile, Jacobs School faculty members gave examples of how AI could be used for other robotics applications.

Michael Yip, a professor in the Department of Electrical and Computer Engineering, made the case for using humanoid robots for telemedicine. As waiting rooms fill up, doctors get increasingly burned out, and surgeries take longer to schedule or get cancelled, humanoid surgical robots offer a solution, Yip argues. Remote surgeons or nurse operators could be brought in to handle cases in areas that are being hit hard by labor shortages, and could provide reprieve to many underserved communities that lack timely and high-quality care. Future humanoid robots could be automated to help take over assistive tasks, working alongside human staff to allow patients to get care quickly and efficiently and reduce the stress and burden teams face today. This would be a game-changer, Yip said.

Falko Kuester, a professor in the Department of Structural Engineering and Computer Science, showed how machine learning-based AI can be used to detect natural hazards and mitigate the damage they cause. Kuester is one of co-principal investigators behind ALERTCalifornia, a public safety program that manages an ever-growing network of than 1,200 natural hazard monitoring cameras across the state. Using the cameras and their associated AI tool, first responders with the California Department of Forestry and Fire Protection (CAL FIRE), and other federal, state, and local government agencies can rapidly confirm that a fire has started, quickly scale fire resources, support evacuations through enhanced situational awareness and monitor fires through containment.

Xiaolong Wang, a professor in the Department of Electrical and Computer Engineering, showed how his team built humanoid robots that mimic expressive human movements for general-purpose, whole-body manipulation while maintaining balance on varied terrains. Beyond whole-body motion and control, Wang’s team has made remarkable progress in robotic dexterity. They have designed robots capable of sorting, packing and folding, as well as a href="https://today.ucsd.edu/story/robotic-hand-rotates-objects-using-touch-not-vision" target="_blank">robotic hands that can handle delicate objects by touch alone, without relying on vision.

Nick Gravish, a professor in the Department of Mechanical and Aerospace Engineering, made the case for integrating computational intelligence and mechanical intelligence. His research group builds robots that use body mechanics to function. Most recently, Gravish’s team developed a gripper based on measuring tape mechanics. Gravish’s team also builds robots modeled on biological processes and animals. He collaborated with researchers at the Georgia Institute of Technology to design and build robots that helped achieve a major breakthrough in understanding how insect flight evolved.

Audience members also got to hear from Robert Webster, senior associate provost for commercialization and technology transfer at Vanderbilt University on commercializing medical robots, and from Dimitry Fisher, senior vice president of data science at Aicadium on industrial AI.

The forum concluded with an open house of robotics collaboratories at Franklin Antonio Hall.

Learn more about research and education at UC San Diego in: Artificial Intelligence