Seite - 117 - in Proceedings - OAGM & ARW Joint Workshop 2016 on "Computer Vision and Robotics“

TowardsAgriculturalRobotics forOrganicFarming ∗ GeorgHalmetschlager1, JohannPrankl1, andMarkusVincze1 FacultyofElectricalEngineering,Automationand Control Institute, Vision for Robotics Laboratory,ViennaUniversityofTechnology,A-1040Vienna,Austria. lastname@acin.tuwien.ac.at Abstract In big scale agricultural farming complex machines with advanced technology shape already the daily routine. In opposite, the field of organic farming is still characterized by multiple manual tasks that include heavy labor. Our vision is that the fields of automation and robotics offer the necessary technology to lift the burden of back-breaking work off the worker’s shoulders. Hence, we propose a scalable and modular agricultural robotic concept that advances farming to the next higher technol- ogy level. Weprovidea low-costandflexibledesign inorder torealizedifferentautonomousapplica- tions, specialized for light weight agricultural work. As proof of concept the proposed configuration is integrated and validated as the experimental platform FRANC. All experiments are performed in real-lifeoutdoorscenariosasvegetablefields thataresowedorplanted inrowstructures. Therefore, weutilizea localnavigationsystembasedonaself-parameterizingcroprowdetection, thatenablesa local,adaptable,andGPS-independentnavigation. The tests showthat thehardwareandsoftwareof the designed system is able to handle rough terrain, offers a high maneuverability, and is adaptable todifferent row-structures. 1. Introduction Within the last decades new automation technologies, industrial robots and sophisticated automa- tion machineries entered the food production chain and led to a higher efficiency and increased the productivityof theharvestingprocess. Sensors and software that transform classic agricultural machineries into semi-autonomous systems are already available on the market [11]. We belief that robotics has the ability to advance this semi- autonomous systems to the next higher technological level and promises to answer the question how theproductionchaincanbefullyautomatedineachsinglestepofthefoodproduction,startingalready at the cultivation of the crops. Therefore we developed a scalable, and modular agricultural robotic systemssuitable tobetter support light-weight agriculturalwork. Asstatedby[17],onewayto increase theeconomicefficiency infuturecropproductionmaybedone better and cheaper with a swarm of small machines than with a few large ones. By the means of our modular concept re-designing existing solutions can be avoided and it becomes possible to enhance existing solutions with robotic modules, for instance a conventional finger cultivator can be turned intoa robotbyattaching the respective robotmodule. ∗ThisworkwasfundedbySparklingScienceaprogrammeof theFederalMinistryofScienceandResearchofAustria (SPA04/84 -FRANC). 117