Page 10 - Read Online

Page 154 Ji et al. Intell Robot 2021;1(2):151-75 https://dx.doi.org/10.20517/ir.2021.14

Figure 1. Review framework.

of applications, and the deep learning models adopted. How to apply these deep learning techniques for
applications and the advantage and limitations of deep learning methods are discussed before some possible
further research areas are outlined as well. Illustrative case studies are also included to show the practical
considerations of applying deep learning methods to rail track condition monitoring. This review
framework provides a balance between deep learning methods and their application to rail track condition
monitoring. The scope is only about condition monitoring of rail tracks; we do not review other rail
components such as rolling stocks or pantographs. Our studies also focus on the deep learning methods
instead of more broadly artificial intelligence or machine learning approaches. The review framework caters
to the needs of both practitioners who need to solve operational issues and researchers who might have
more interests in the methodologies.