This is not your casual weekend read.

We are hearing over and over this year that artificial intelligence (AI) is finally “taking over the world.” Maybe or maybe not. When I received the request to review this paper, I felt the need to get educated for that apocalypse.

This is a paper dealing with solutions for controlling/tracking nonholonomic mobile robots. For those not so familiar with this domain, that means there are more dimensions of freedom to the movement than can be controlled. For example, a car moves along the direction of the wheels, but there can be movement in more dimensions. The authors point out that many prior studies rely on modeled dynamics, which are realistically unavailable. One solution is to introduce adaptive laws through neural networks to estimate the unknown upper bounds of such unmodeled dynamics and external disturbances. These adaptations are used online such that modeled dynamics can be applied.

The paper is suitable for readers with an interest in AI and a strong mathematics background. It uses a lot of advanced mathematics for the approximation of strongly nonlinear and complex systems in neural networks.