No abstract

No abstract

When developing control architectures for physical human-robot interaction, it is often necessary to use a model of the human operator. This paper presents a comparative study of both direct and inverse models of the human arm at the elbow joint, relating the force at the hand with the arm angle and vice versa. Specifically, models of integer and fractional (commensurable and non-commensurable) nature are identified from the experiments. Likewise, for comparison purposes, neural networks models are also obtained. Taking into account their parameter variability, it is shown that fractional models are more adequate to describe human arm behaviour; they are simpler, more exact and with less parameter uncertainty.

No abstract