The urban lawns are frequently composed by different grass species to combat some problems of water scarcity and diseases. In order to maintain these lawns, high amount of water is required. Nowadays, smart cities can be understood as a new concept of city that includes, among others, efficient distribution of energy, water, and other resources by using technology. In these cases, the main challenge is to try to estimate the necessary amount of water for irrigation and the phytosanitary uses without wasting water. In this paper, we propose a method to identify the percentage of grass coverage in lawns to deduce the grass productivity and estimate the most accurate quantity of water to ensure a good production of grass. The system is based on a Smart Autonomous Vehicle (SAV) controlled by an Arduino Mega 2560. It also contains an array of 120 colour sensors used to gather the data. The selected colour sensor is a TCS3472. With these sensors, we obtain the RGB histograms of the lawns. For these experiments, we have several lawn parcels of 1.5 x 1 m. From these, a matrix of 150 x 100 RGB values is obtained. After processing the green values of matrix, we have observed a correlation between the level of coverage and these values. The grass coverage is related with values of brightness between 40 and 60 which allow us to classify the lawn as a function of its coverage and the irrigation needs.