Robotic planning tool accounts for human negligence

A new algorithm could make robots safer by making them more sensitive to human inattention. In computer simulations of packaging and assembly lines where humans and robots work together, the algorithm developed to account for human inattention improved safety by about 80% and efficiency by about 38% compared with existing methods.

The work is reported in IEEE Transactions on Systems, Man, and Cybernetics: Systems.

“There are a large number of accidents that happen every day due to negligence – most of them, unfortunately, are due to human error,” said lead author Mehdi Hosseinzadeh, an assistant professor in the School of Mechanical and Materials Engineering at Washington State University.

“Robots act as expected and follow the rules, but humans often don’t. This is the most difficult and thorny problem.”

Robots working alongside humans are becoming more common in many industries, where they often work together. Many industries require humans and robots to share a workspace, but repetitive, tedious work can cause people to lose focus and make mistakes.

Most computer programs help robots respond when an error occurs. These algorithms may focus on improving efficiency or safety, but they don’t take into account changes in the behavior of the people they work with, Hosseinzadeh said.

As part of their efforts to develop a blueprint for robots, the researchers first worked to quantify human negligence, looking at factors such as how often a human ignores or misses a safety alert.

“We defined the degree of inattention, and the robot observed the human’s behavior and tried to understand it,” he explained. “The notion of degree of inattention is something new. If we know which human is inattentive, we can do something about it.”

Once the robot identifies reckless behavior, it is programmed to change the way it interacts with the human who acts in that way, working to reduce the risk that the person could make a mistake in the workplace or get hurt.

So, for example, the robot can change the way it handles its tasks to avoid disturbing the human. The robot constantly updates the level of carelessness and the changes it observes.

The researchers tested their project using a computer simulation of a packaging line with four people and a robot. They also tested a simulated collaborative assembly line where two humans would work together with a robot.

“The main idea is to make the algorithm less sensitive to the behavior of careless humans,” Hosseinzadeh said. “Our results revealed that the proposed system has the ability to improve efficiency and safety.”

After running a computer simulation, the researchers plan to test their work in the lab with real robots and humans, and eventually in field studies. They also want to quantify and account for other human characteristics that influence workplace productivity, such as human rationality or awareness of danger.