Using the Internal Model Hypothesis to Model Human Motor Control in Multi-Domain Simulations

Abstract

There is a trend in human-machine interaction towards a more physical interaction between humans and machines. Examples are shared control systems such as lane keeping assistants or human-robot collaboration. Thereby, the interplay of the sensorimotor control loops of the human operator with the machine plays an important role. State of the art literature suggests that this interplay should be considered already in the design process of such systems. Therefore, the realistic modeling of the human motor control is crucial for a human model for physical human-machine interaction. The current state of research in neuromechanics suggests that humans use so-called internal models of their limbs and environment, located in the central nervous system, to perform accurate movements despite slow reaction times. This theory is called the internal model hypothesis. This paper presents an approach to implement the internal model hypothesis in a dynamic human model using inverse models. The inverse models are derived using modified Newton-Euler equations. Finally, the validity of the chosen approach is demonstrated by comparing numerical simulations with experimental data using the example of a human-steering wheel interaction. The numerical simulations show a similar qualitative behaviour compared to the mean subject trajectories. Furthermore, the simulated trajectories are within the standard deviation of the experimental data for most of the time.