Meet the Georgia Tech robot that is learning how to help people get dressed

So far, the machine can only help ease the first arm onto an experimental subject. But it learned to do so safely and effectively by running more than 11,000 physics-based simulations of the process.

“Some of those attempts were flawless,” according to a statement from the university. “Others were spectacular failures–the simulated robot applied dangerous forces to the arm when the cloth would catch on the person’s hand or elbow.”

Still, running those tests on simulated humans rather than real-life volunteers helped avert danger from the robot, which can now successfully dress actual people’s arms, even in cases where the gown initially gets caught on the elbow. Fully dressing someone in the gown will require more research, but it could one day be an invaluable service to people with disabilities that make putting a gown on unassisted difficult, the researchers say. The robot uses the forces it experiences as it dresses a human, rather than computer vision, to guide it.

To conduct the test, the researchers used a PR2, a robot developed by the Californian robotics company Willow Garage. The research team was led by Zackory Erickson, a Georgia Tech PhD student. “People learn new skills using trial and error,” Erickson said in a statement. “We gave the PR2 the same opportunity. Doing thousands of trials on a human would have been dangerous, let alone impossibly tedious. But in just one day, using simulations, the robot learned what a person may physically feel while getting dressed.”

According to the release, the research was supported in part by an Award from the National Science Foundation, AWS Cloud Credits for Research, and a research traineeship program. Next up, the researchers plan to present their work at the International Conference on Robotics and Automation in Australia, which takes place later this month.