Anticipating partner's movements - news

Anticipating partner’s movements – news

December 07, 2022

Imagine an industrial robot strong enough to lift an engine block and perceptive enough to safely reposition and rotate that piece of metal while a human attaches it to the vehicle or screws on additional parts.

Manufacturers such as the Ford Motor Company have seen this potential use of robots as a way to add flexibility to assembly lines and reduce the need for expensive reconfigurations. After three years of work, a team from Carnegie Mellon University Robotics Institute (RI) recently demonstrated the integration of the realization and action needed to make it a reality.

“We want robots to help humans perform tasks more efficiently,” said Ruixuan Liu, Ph.D. Robotics student. This means developing systems that allow a robot to track the progress of a human partner on a collection task and anticipate the partner’s next move.

in proof For Ford researchers in October, the team used a wooden box attached to a robotic arm to show how these systems work. For the demonstration, a human had to insert blocks of different shapes into the corresponding holes on different sides of the box. The robot, in turn, rotates the box so that the human can quickly insert a block into the appropriate hole.

The robot was able to assist in the task regardless of which order the human chose for the blocks and how the blocks were arranged. They also internalized the behaviors of different human partners.

“The traditional approach is to first see which block a person chooses,” he said. Changliu Liu, an assistant professor at RI who led the project. “But this is a slow approach. What we’re doing is actually anticipating what the human wants so the robot can act proactively.”

This method requires the robot to change its predictions based on differences in behavior between workers or even in the behavior of the worker itself over time.

“Humans have different ways of moving their arms,” ​​Liu said. “Sometimes they may put their elbow on the table. And when they get tired, they tend to lower their arm a little more.”

Teaching a robot to understand different behaviors usually requires collecting a huge amount of data about human behavior. Liu’s group found ways to reduce training so that they didn’t need to define a lot of rules or do a lot of training beforehand.

Accurate prediction of intent is critical to ensuring the safety of the human factor. Industrial robots usually operate in closed cages to prevent human injury. In this demo, a human and a robot occupied the same workspace.

Their proximity required the robot to predict in real time where the human was likely to move and then dynamically allocate a protected area around the human within the shared space.

“This is a completely different way of doing things,” said Greg Linkowski, a Ford robotics research engineer who watched the demo. “It was an interesting proof of concept.”

Liu and her group’s work was supported by the Ford University Research Program. Liu and Linkowski said discussions are underway to continue the research by expanding it to multiple robots and multiple workers. It’s impossible to say if the engine-assembly task that inspired the demo will eventually be carried out, but Linkowski noted that automakers have a huge interest in taking advantage of the increased robotic capabilities.

“That was just one possibility,” Liu said. While the robot was serving as an assistant in the most recent demonstration, Liu said, it’s easy to imagine applications where both humans and robots are actively assembling the device, “where the human does the more complex assemblies and the robot does the easier ones,” Liu said. .

#Anticipating #partners #movements #news

Leave a Comment

Your email address will not be published. Required fields are marked *