The specific problem which initiated my research was the study of the clustering properties of galaxies as revealed by large redshift surveys, a context in which concepts of modern statistical physics (e.g. scale-invariance, fractality, etc.) found ready application. We have then considerably broadened the range of problems in cosmology which we have addressed, treating in particular more theoretical issues about the statistical properties of standard cosmological models and attempting to propose some theoretical models. What is common to all this research is that it is informed by a perspective and methodology which is that of statistical physics. Beyond my close collaborators in this activity (Luciano Pietronero, Michael Joyce, Andrea Gabrielli) I have collaborated with both well-known statistical physicists, as Phil Anderson (Princeton), Jean Pierre Eckmann (Geneva), and Joel Lebowitz (Rutgers) and cosmologists as Ruth Durrer (Geneva), Martin Lopez Correidoira (IAC Las Canarias),  and Helene Di Nella-Courtois (Lyon). I can say that, beyond its specific scientific content, this activity has an underlying key-point: such interdisciplinary research is an exciting playground for statistical physics, and one which can bring new and useful insights into cosmology and astrophysics