Seismic surface waves analysis has become a relatively standard tool for a wide range of applications in Earth sciences, mainly to evaluate soils and rocks shear properties and to image near-surface heterogeneities. The underlying theories, the data processing workflows and interpretation methods are quite similar whatever the scales of interest (from global seismology to ultrasonic non-destructive testing). In each domain, the acquisition techniques are rapidly evolving with the development of innovative sensors enabling: wireless 3-components surveys; dense arrays; distributed measurements; continuous monitoring etc. In the meantime, most operational analysis tools remain based on the approximations that the wavefields mainly involve plain waves and that the probed media are elastic and stratified with smooth geometries compared to recorded wavelengths. Even if numerical modelling and data processing methods have made huge progresses in the last ten years (for instance with the democratisation of massively parallel computing platforms), addressing theoretical and methodological issues related to near-surface seismic prospecting is difficult as soon as the 3D nature of the Earth and its high degree of heterogeneity have to be taken into account, which is quite frequent, though! More than 15 years ago, we suggested the use of laboratory physical modelling and laser-Doppler small-scale surveys to complement numerical studies and to perform experimental benchmarks --and sometimes validations-- of surface-wave processing and inversion techniques. In this manuscript we show how, since then, we developed this approach on 3D physical models of increasing degrees of complexity, in order to study seismic-wave propagation in unconsolidated and porous media, so as to target near-surface environmental, geological and geotechnical applications. We also provide guidelines for practitioners regarding the operational use and limits of surface-wave methods, e.g.: their ability to track compaction and rigidity anomalies in natural or artificial soils; the possible interpretation of near-surface seismic measurements in terms of spatio-temporal variations of water content. We also suggest alternative approaches to be addressed in the near future according to recent instrumental and computational developments cited supra.