Scientists simplify model for human behavior in automation

October 14, 2016, Chinese Association of Automation

Human unpredictability is a problem in the automated human-machine systems people use every day. Scientists from Nanjing Institute of Technology's School of Automation in China and the University of California, Merced's School of Engineering partnered to find a programming solution for erratic human behavior.

"In a human-machine control system, the participates in the control process," said Jiacai Huang, a professor at the Nanjing Institute of Technology's School of Automation and the study's lead author. Human operator behavior includes not only skilled control tasks, but also instinctive and emotional reactions. "[An] accurate mathematical model of human operator behavior provides criteria to the controller design and systems analysis."

The scientists published the theory and experimental evidence for a new human behavior prediction method in the IEEE/CAA Journal of Automatica Sinica.

Control systems, such as a car's cruise control feature, are based on specific parameters with limited variables. On a hill, the cruise control will automatically accelerate to maintain speed against gravity. For most other variables, such as another car slowing unexpectedly, the system depends on the driver to apply the brakes.

The system doesn't make decisions; it simply reacts to the input. Yet the inability to perfectly predict human operator behavior hinders advanced system design. Researchers have analyzed how best to describe in machine systems since the mid-1940s. Currently, scientists use several equations to account for every potential outcome.

Credit: Chinese Association of Automation

The researchers' proposed model is based on fractional order calculus, in which multiple outcomes can be considered within the same equation.

"[Our] model for human operator behavior has many advantages, such as simple structure with few parameters, [all] with clear physical meaning," said Huang. "More importantly, the proposed new model gives a unified, formalized description for the human operator behavior."

The human operator is modeled as a part of the system rather than an addition to the system, as traditional modeling does. To experimentally test this idea, the scientists created a closed-loop system in which a person followed a target on the screen with a steering wheel. Compared to traditional mathematical models, the proposed method was a better fit for how the human operator actually behaved. "The human operator is a complex system, and many aspects of the human brain and behavior have the characteristics of a fractional order system," Huang said.

The researchers plan to study how their human operator model could improve advanced systems, such as autopilot in planes or for robotic use in surgeries.

Explore further: Continuous adaptation makes for more natural interactions between robots and humans in shared tasks

More information: Fractional order modeling of human operator behavior with second order controlled plant and experiment research, DOI: 10.1109/JAS.2016.7508802 , http://ieeexplore.ieee.org/document/7508802/

Related Stories

The motorcycle of the future won't need a rider

September 12, 2016

Microengineering student Eric Unnervik has developed a miniature self-balancing motorcycle. His aim is to one day produce an entirely autonomous motorcycle that can outperform human riders.

Your next nurse could be a robot

October 5, 2016

The nursing assistant for your next trip to the hospital might be a robot. This is the implication of research recently published by Dr. Elena De Momi and colleagues in the open access journal Frontiers in Robotics and AI ...

Recommended for you

1 comment

Adjust slider to filter visible comments by rank

Display comments: newest first

Hyperfuzzy
not rated yet Oct 24, 2016
Not sure i get it, it looks like an object that seeks best mimic. I did not see an analytical model of the car's safe zone; something like a computed ellipsoid about the car. The computer can update its expectations vs actuality continuously, in fact it would be exception control. Using possibility of a computed unsafe space when meeting other drivers would be simpler if all cars actually talked, thus eliminating human error. The last calculation with possibilities of a human at the wheel has difficulty avoiding an attacker. Yes its better than human but with present road design, especially in Mo.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.