A new algorithm has the potential to enhance robot safety by increasing their ability to recognize when humans are not paying attention. In virtual simulations of packaging and assembly lines, where robots and humans collaborate, this algorithm has shown an improvement in safety of up to 80% and efficiency of up to 38% compared to current techniques.
A new algorithm has the potential to enhance robot safety by increasing their ability to recognize when humans are not paying attention.
In virtual simulations of packaging and assembly lines, where robots and humans collaborate, this algorithm has shown an improvement in safety of up to 80% and efficiency of up to 38% compared to current techniques.
This research is detailed in IEEE Transactions on Systems Man and Cybernetics Systems.
Lead author Mehdi Hosseinzadeh, an assistant professor at Washington State University’s School of Mechanical and Materials Engineering, stated, “Many accidents occur daily due to carelessness, mostly arising from human mistakes. Robots follow directives; however, humans frequently fail to adhere to guidelines, creating significant challenges.”
In various industries, collaboration between humans and robots is on the rise, with both often sharing workspace. Tedious and repetitive tasks can result in a loss of focus, leading to errors. While most existing algorithms help robots respond to mistakes once they happen, they tend to prioritize either efficiency or safety without accounting for the fluctuating focus of their human counterparts, according to Hosseinzadeh.
The researchers aimed to quantify human inattentiveness by analyzing factors such as the frequency with which a person overlooks or misses safety warnings.
“We established a definition for carelessness, enabling the robot to observe and understand human behavior,” he explained. “The concept of carelessness levels is innovative. Identifying inattentive individuals allows us to take appropriate action.”
When a robot detects careless behavior, it adjusts its interactions with the inattentive person to minimize the likelihood of workplace errors or accidents. For instance, the robot might modify its task management to avoid obstructing the human worker. Moreover, it continually updates its understanding of the carelessness levels and any observed changes in behavior.
The research team evaluated their algorithm using computer simulations that involved a packaging line with four humans and a robot, as well as a simulated assembly line featuring two humans working alongside a robot.
“The central concept is to make the algorithm less reactive to the actions of inattentive humans,” said Hosseinzadeh. “Our findings indicate that this approach can enhance both efficiency and safety.”
After success with the simulations, the researchers intend to validate their findings in a lab setting with actual robots and human workers, progressing to field studies in the future. They also plan to explore additional human characteristics, such as rational decision-making and hazard awareness, which could impact workplace effectiveness.
This research was supported by the National Science Foundation, with co-authors Bruno Sinopoli and Aaron F. Bobick from Washington University in St. Louis.
