Hyperspectral imaging is a trending technology in recent years. The growing interest in this technology is due to the richer information that spectral data carries compared to classic RGB imaging. To study the potential of this technology for Smart Viticulture, we have built a robotic platform equipped with three hyperspectral cameras and other classical sensors used in autonomous mobile robotics to collect data in an experimental vineyard and at a commercial vineyard. This data is targeted to grape bunches counting and volume estimation, thus computing yield predictions. The ongoing experimentation aims to use hyperspectral images to estimate the sugar and anthocyanin grape concentration and detect leaves exposed to biotic/abiotic stresses.