Researchers at Carnegie Mellon University’s Robotics Institute (RI) are working towards developing a robot-assisted dressing system to assist individuals who need help getting dressed. The system aims to address the diverse range of clothing options and motor function capabilities of different individuals.

The researchers employed artificial intelligence and reinforcement learning to teach the robot how to manipulate clothing and dress people. They used a simulation to train the robot, considering the properties of the clothing material when transferring the learned strategies to the real world. The team used a high-dimensional representation called a segmented point cloud to perceive the current state of the clothing and its interaction with the human’s arm.

In a human study involving 510 dressing trials across 17 participants, the system was able to successfully dress the participants in most cases. For the majority of participants, the system fully pulled the sleeve of each garment onto their arm, dressing 86% of the length of their arms on average.

The researchers also emphasized the importance of safe human-robot interaction, taking measures to ensure the robot applied gentle force and avoided actions that could cause discomfort or compromise safety.

Future research aims to expand the capabilities of the system by enabling it to dress a person in a jacket or pull a T-shirt over their head, as well as explore more advanced manipulation skills like buttoning or zipping. The researchers also plan to perform observational studies within nursing facilities to gain insight into the diverse needs of individuals and further improve the system.