Safe work is not just about processes, but about context – understanding the work environment and conditions and being able to predict what other people will do next. The new system allows robots to have this level of awareness so they can work in parallel with humans on assembly lines more efficiently and without unnecessary interruptions.
Instead of only estimating the distance between himself and his co-workers, the human-robot collaboration system can identify each employee he or she works with, as well as a person’s skeletal model that is abstract from body volume, says Hongyi Liu, a researcher at KTH’s Royal Institute of Technology. institute. Using this information, the context-aware robot system detects the employee’s position and even predicts the next position. These abilities provide the robot with a context that needs to be aware of when interacting.
Liu says the system works with artificial intelligence, which requires less computing power and smaller data than traditional machine learning methods. Instead, it is based on a form of machine learning called transfer learning – which utilizes the knowledge developed through training before adapting it to a business model.
The study was published in the latest issue Robotics and computer-integrated manufacturing, and was written by Professor Lihui Wang of KTH.
Liu says the technology is ahead of International Organization for Standardization (ISO) requirements for the safety of collaborative robots, so the introduction of the technology would require industrial action. But context awareness offers better efficiency than the one-dimensional interaction that employees now experience with robots, he says.
“According to the ISO standard and technical specifications, when a human approaches a robot, it slows down and if he gets close enough, it stops. If the person moves away, it continues. It’s a fairly low context awareness,” he says.
“It jeopardizes efficiency. Production slows down and humans can’t work closely with robots.”
Liu compares the context-aware robot system to a self-driving car that detects how long the brake light has been red and predicts movement again. Instead of braking or shifting, it begins to adjust its speed by cruising toward the intersection, saving brakes and transmission.
Experiments with the system showed that in context, a robot can operate safer and more efficiently without slowing down production.
In one system test, someone’s hand unexpectedly blocked the path of the robot’s arm. But instead of stopping, the robot adapted – it predicted the future trajectory of the hand and the arm moved around the hand.
“This is not only safety from a technical point of view in avoiding collisions, but is able to identify the context of the assembly line,” he says. “This gives an extra layer of security.”
The study was a follow-up to the Symbiotic Human Robot Collaborative Assembly project, which was completed in 2019.