Research On Visual Positioning Method And Application Of Power Line Patrol For Rotor Flying Robot

With high mobility and the feature of vertical take-off and landing and hovering, Rotor Flying Robot can undertake dangerous tasks and get information on high-risk environments task in replace of human being. As advances in sensor and calculation methods, leading to cost reduction in producing Rotor Flying Robot, its performance and functionality has been expanded in many ways. In addition to the original applications in military defense, it has been applied in all aspects of human life, which greatly broadens our horizon and strengthens our ability. Research of the perception algorithms and methods of control needs experiments in indoor environment, while under such circumstance where GPS cannot be used, aircraft usually fail to get the location information. For this reason, this thesis studies the external visual and onboard visual positioning method, base on which the application of visual inspection in power line patrol has been research.Firstly, this thesis presents a systematic study of the application background of Rotor Flying Robot covering an outlines if well-known suppliers both at home and abroad, as well as current domestic and overseas research status and applications prospect of vision technology in this field.This thesis studies positioning methods for rotor flying robot based on external vision. Utilizing RGB-D sensor, we built the system which is suitable for Rotor Flying Robots indoor. Our system has the advantages of low cost and fast speed which have laid a solid foundation for the latter project research on power line inspection method and Rotor Arm control. We have designed a visual tracking algorithm for the target of Rotor Flying Robot, which calculates the world coordinates of the target through color image and depth data from RGB-D sensor. The proposed system makes it possible for the aircraft autonomous to maneuver hovering and displacement, and what's more, attributes to the completion of the experiment in which the aircraft follows a particular track, and the precision that Rotor Flying Robot can hover is less than 15mm.This thesis studies the localization method based on the ground mark with an onboard camera. The ground marking system is composed of a few simple geometric shapes, and we try to classify the marks by a graphics classifier called Bagging. We first extract the relevant points target in image coordinates, and then use EPnP pose solver algorithm to calculate rotor flight robot position based on the ground coordinate system. The results, in comparison to EPVS positioning result, displays deviation less than 10mm. We also discussed about the capability of visual positioning under the situation that the marks cannot be fully recognized, the result shows our method is effectiveness and stable.Against powerline imaging characteristics of urban traffic road, we have proposed the power line detection method based on Hough Transform and Kmeans clustering. Experiments show that the proposed image processing program can well detect the power line in the image of urban traffic road, and accurate position the power line and calculate the parameters. We have also studied the image servo method based on power line imaging for rotor flying robot. The simulation results show that this method can well complete the alignment work.