Feasible strongUAVstrong Path Planning Using Genetic Algorithms and.pdfVIP

Feasible UAV Path Planning Using Genetic Algorithms and B´ezier Curves Douglas Guimar˜aes Macharet, Armando Alves Neto, and Mario Fernando Montenegro Campos Computer Vision and Robotic Laboratory (VeRLab) Department of Computer Science, Universidade Federal de Minas Gerais Belo Horizonte - MG - Brazil {doug,aaneto,mario}@dcc.ufmg.br Abstract. With the growing in the use of UAVs (Unmanned Aerial Ve- hicles), it is necessary to develop techniques that allow the generation of feasible paths for these vehicles. These paths take into account the nonholonomic constraints intrinsic to UAVs, such as minimum curva- ture, minimum torsion and maximum climb (or dive) angle. Thus, this paper proposes the use of genetic algorithms to generate paths for these vehicles in the three-dimensional space, using B´ezier curves with several advantages. We consider all these three constraints in order to generate a feasible path for a small fixed-wing aircraft with severe limitations. We show results for this vehicle. Keywords: Path planning, UAVs, B´ezier curves, Genetic algorithm. 1 Introduction Nowadays, a great effort is being invested to increase the ability of mobile robots to make their own decisions. Especially on issues linked to its mobility, reducing or eliminating altogether the need for human intervention on robot’s assigned tasks. Among the various issues involved, one well known problem is posed by the mobile robotics basic questions related to navigation: “How does it get to a given goal?”. The question is directly linked to the strategy being used by the robot to safely achieve a goal position. Several strategies to obtain a p