Why human plays an important role in control loop?
Humans’ unique abilities, such as adaptive behavior in dynamic environments, social interaction, and moral judgment capabilities, make humans essential elements of many control loops, operating in close collaboration with autonomy. Compared to human control, autonomy provides higher computational performance and multi-tasking capabilities without fatigue, stress, or boredom. Thus, we need to use both human and autonomy in the control loop.
Why is it important to analyse human behavior in the control loop?
The probability of human error causing system failure is higher. Moreover, humans may have anxiety, fear, and unconsciousness during operations. In the tasks requiring increased attention and focus, humans may tend to provide high gain control inputs, which can cause undesired oscillations, such as pilot-induced oscillations (PIO).
How to model human behavior?
Reliable human mathematical models are required to develop safe control mechanisms and better realize and understand human control actions and limitations. Human behavior in the control loop is due to 1) sensory model, which consists of visual, vestibular, proprioceptive, etc. 2) neuromuscular model, which is considered as the relation between brain signal and muscle movements, 3) control theoretic model, which is related to the decisions taken by human to stabilize the system or compensate disturbances, etc.
Adaptive human model
Inspired by humans’ ability to adapt to changing environments, the proposed adaptive human model mimics this ability despite input bandwidth deviations and plant uncertainties. The proposed human pilot model structure is based on the model reference adaptive control, and the adaptive laws are obtained using the Lyapunov Krasovskii stability criteria.
Experimental setup
A real-time experimental setup to examine whether the proposed adaptive human model mimics the human behavior in the presence of uncertainty is provided. A Logitech Extreme 3D Pro joystick is used, which transfers the participants’ data to MATLAB. SIMULINK is also used to prepare the display, where participants should follow a target.
Model validation by statistical analysis
Model validation is done by comparing the adaptive human model and participants’ data. Statistical analysis consists of confidence interval calculation, hypothesis test, and power analysis is conducted to measure the predictive power of the proposed model. You can access the participants’ data by clicking on the following green button.
Related references
Shahab Tohidi and Yildiray Yildiz, “A control theoretical adaptive human pilot model: Theory and experimental validation,” IEEE Transactions on Control Systems Technology, 30 (6), 2585-2597, 2022.
Shahab Tohidi and Yildiray Yildiz, “A control theoretical adaptive human pilot model: theory and experimental validation,“ arXiv preprint arXiv:2007.10216, 2020.
Shahab Tohidi, and Yildiray Yildiz, “Adaptive human pilot model for uncertain systems,” European Control Conference (ECC), 2938-2943, 2019.