Towards Smart Farming and Sustainable Agriculture with Drones Paolo Tripicchio and Massimo Satler Gustavo Stefanini Advanced Robotics Research Center Scuola Superiore Sant’Anna Email:p.tripicchio@sssup.it,m.satler@sssup.it Giacomo Dabisias Emanuele Ruffaldi and Carlo A. Avizzano PERCeptual RObotics Lab Scuola Superiore Sant’Anna Email: g.dabisias@sssup.it, e.ruffaldi@sssup.it Abstract—The use of drones in agriculture is becoming more and more popular. The paper presents a novel approach to dis- tinguish between different field’s plowing techniques by means of an RGB-D sensor. The presented system can be easily integrated in commercially available Unmanned Aerial Vehicles (UAVs). In order to successfully classify the plowing techniques, two different measurement algorithms have been developed. Experimental tests show that the proposed methodology is able to provide a good classification of the field’s plowing depths. I. I NTRODUCTION The continuous growth of the world population together with the lowering of resources at disposal pose the problem of smart usage of resources. This is very important especially in the field of food production and soil exploitation. The common methods used in agriculture to analyze and assess the correct production and usage of resources employs optical and multispectral techniques applied to photos captured from satellites. These techniques allow to assess the health state of farmings; for instance the light absorption from the leafs displays the presence of chlorophyll. This is a critical and important phase since the results of these phase will affect the decisions of interventions on the feeding of the soils, the pro- tection from insects/fungi or if other countermeasures should be taken. The more frequently this kind of analysis is done, the more responsive and thus accurate the countermeasure will result. On the other hand this activity is time consuming if held by hand and satellite-time dependent if done by this kind of technology. In the last years, there is a growing interest in the use of autonomous techniques for inspecting the health state of farming. Robotics jumped into this field providing interesting and effective solutions to several phases like harvesting or the plowing [1]. Compared with the satellite technology, the use of drones in agriculture and in smart farming is very effective due to the fact that unmanned aerial vehicles (UAV) can give farmers a bird’s eye view of their fields still remaining close to the terrain and so providing more precise evaluations. In particular, the use of drones does allow the opportunity to get an overall survey of the area and make a better use of farmer time, rather than just making him/her walk out blindly into a field that could be taller than his/her head, hoping that he/she stumble across any of the problem areas that might be in the field. Fig. 1. Pre-programmed navigation trajectory for the soil assessment in the APM Planner open-source software. Recently, to protect the natural ecosystem of farming fields, the Italian Tuscany region has given subsidies for the lavoration of small fields, giving prizes to farmers which will reduce the fields plowing depth by changing their plowing techniques. This direction looks promising and will be taken over probably by other regions and countries in the next future. To be able to assess directly the effective usage of the soils, satellite images could not be sufficient or should be validated at least from a closer inspection method. For this purpose, our work aims at developing a system capable of analyzing the soil condition with a rapid flight. The idea is to approach a correlation between radar (or satellite) acquired parameters and soil roughness values obtained from RGB-D cameras or laser scanners. The paper is divided as follows: next section will introduces a brief description of the state of the art solutions. Section III will give a general overview of the system architecture. In section IV the data acquisition method is presented and analyzed in section V. Section VI shows the results of the data analysis procedure and finally section VII reports the paper statement together with conclusions and future development of the project. II. STATE OF THE ART Optical satellite images require specific time slots to be acquired and usually are quite expensive; for this reason a