A humanoid robot trained to effortlessly learn and perform a variety of expressive movements, including simple dance step sequences and gestures such as waving, high-fiving, and hugging while maintaining a steady gait across a variety of terrain has been shown to positively impact human-robot interactions in work environments.
That’s according to a study by engineers at the University of California San Diego, presented at the Robotics: Science and Systems conference, taking place July 15-19 in Delft, the Netherlands. The increased expressiveness and agility of this humanoid robot paves the way for improved human-robot interactions in environments such as factory assembly lines, hospitals and homes, where robots could operate safely alongside humans or even replace them in dangerous environments such as laboratories. “Through expressive and more human-like body movements, we aim to build trust and showcase the potential for robots to coexist harmoniously with humans,” said Xiaolong Wang, a professor in the Department of Electrical and Computer Engineering at UC San Diego’s Jacobs School of Engineering.
“We’re working to help reshape the public’s perception of robots as friendly and cooperative rather than terrifying like Terminator,” Wang continued. What makes this humanoid robot so expressive is that it’s been trained on a wide range of human body movements, allowing it to generalize new movements and imitate them with ease. Like a fast-learning ballet student, the robot can quickly learn new patterns and gestures. To train the robot, the research team used a large collection of motion capture data and dance videos.
Their technique involved training the upper and lower bodies separately. This approach allowed the upper body of the robot to replicate various reference movements, such as dancing and high-fiving, while the legs focused on a steady stepping motion to maintain balance and traverse different terrains.
“The main goal is to show the robot’s ability to do various things while walking from one place to another without falling,” Wang explained. Despite separate training of the upper and lower bodies, the robot operates under a unified policy that governs its entire structure. This coordinated policy ensures that the robot can perform complex upper-body gestures while walking stably on surfaces such as gravel, dirt, wood chips, grass, and inclined concrete paths.
The simulations were first conducted on a virtual humanoid robot and then transferred to a real robot. The robot demonstrated the ability to perform both learned and new movements in real-world conditions. The robot’s movements are currently directed by a human operator using a game controller, which dictates its speed, direction and specific movements. The team envisions a future version with a camera to enable the robot to perform tasks and navigate terrains completely autonomously. The team is now working to refine the robot’s design to tackle more intricate and precise tasks. “By extending the capabilities of the upper body, we can expand the range of movements and gestures the robot can perform,” Wang said.