Smarter Algorithms: How Technology Recognizes Human Errors

A new algorithm is set to revolutionize the way robots work with humans by making them more aware of human inattentiveness. Developed by researchers at Washington State University, this new tool is designed to account for human carelessness, making robots safer and more efficient in shared workspaces.

In a series of computerized simulations of packaging and assembly lines, the algorithm boosted safety by up to 80% and increased efficiency by up to 38%, marking a significant improvement over existing technologies. This research, published in IEEE Transactions on Systems, Man, and Cybernetics: Systems, is a promising step forward for industries where robots and humans collaborate closely.

“Every day, countless accidents occur due to human errors,” explained Mehdi Hosseinzadeh, the study’s lead author and an assistant professor at Washington State University’s School of Mechanical and Materials Engineering. “Robots follow the rules perfectly, but humans don’t always do the same. That’s the real challenge.”

As robots become increasingly common in industries ranging from manufacturing to logistics, safety concerns are growing. Human workers often lose focus during repetitive tasks, leading to mistakes. Most existing algorithms allow robots to react after an error, but few account for the fluctuating behavior of their human coworkers.

The researchers developed an innovative method to quantify human carelessness, looking at factors such as how often workers miss or ignore safety alerts. Once identified, the algorithm adjusts the robot’s actions in real-time, allowing it to avoid errors and keep the workspace safe.

“If we know which humans are inattentive, the robot can change its behavior accordingly,” said Hosseinzadeh. For instance, a robot might change its movement patterns to avoid obstructing a distracted worker. The robot continuously updates its understanding of carelessness based on human behavior.

In simulated tests, the algorithm was put through its paces on a packaging line with four humans and one robot, as well as on an assembly line where two people collaborated with a robot. The results were clear: the algorithm made robots more resilient to human errors.

The next step for Hosseinzadeh and his team is testing the algorithm with real robots in a laboratory setting. Future research will also look into other human traits that impact workplace safety and productivity, such as decision-making and hazard awareness.

Funded by the National Science Foundation, this research offers a glimpse into a safer future for human-robot collaboration, where technology can adapt to human imperfection and make the workplace more secure.