Seite - 118 - in Proceedings - OAGM & ARW Joint Workshop 2016 on "Computer Vision and Robotics“

Figure1: Experimentalplatformand proof ofconceptFRANC. Inthisarticlewepresentourmodularsystemdesignandconceptsformoreflexibleagricultural robots aswell as its realizationwith theplatformFRANC(cf. Fig. 1)asproofofconcept. Ourcontributions are(i)amodularroboticsystemconceptthat(ii)canbeusefortherobotizingofexistingfarmfacilities and(iii) ageneric rowdetectionalgorithmthatdoesnotneedanya-priori information. Moreover,we presentfield trial resultsof its performanceonvegetablefields. 2. RelatedWork Mostof thestate-of-the-artagriculturalautomationsystemsareeither focusedonthe(semi-)automa- tion of big land machines or support the farmer during different field manipulation procedures with additional sensor information [11]. Research groups robotized already “standard platforms” as golf carts or other small scale vehicles to focusonalgorithmsandsensor technologieswithout theneedof there-developmentof thegrounding vehicle [16, 10, 5]. Contrary there are also completely designed robotic systems such as BoniRob which are suitable for highly specialized solutions as occur in the area of precision farming [3]. As described in [3] BoniRob Apps are comparable to the classical implements. These Apps can be di- rectly integrated on the robot. Existing agricultural machineries have to be redesigned if they have to be used be used with the robot. However, we aim to develop a solution which can be used in combi- nation with existing implements and farm facilities, with little or no additional product development needed. The contribution presented in this article is a system design concept that sets out to close the gap of existing agricultural robotic systems: rather than focusing on one large multi-purpose autonomous machine, we offer a flexible solution for cheaper crop row production which might be even more acceptable for smaller farms as it allows robotizing existing machines. We present our contribution in the form of a detailed system description and the results of preliminary field trials. Our results should help other researchers and engineers to solve the existing challenges in agricultural robotics with respect todevelopment of robotic systems for light-weight agriculturalwork. 3. Approach Weapproacha robotic system, includinga rowguidanceanautonomymodule that adaptsby itself to any kind of row organized fields. Our concept includes individually replaceable subsystems that will bepresentedhere. 118