VSN-based Indoor Localization And Path Planning For MAV

The past few decades have witnessed a marked growing heat over micro aerial vehicle(MAV)thanks to its advantageous of being low cost,low power-consumption,light weight and flexible in flight performance.MAV has shown great potential in both military and civilian applications and indoor applications have become a development trend.However,practical indoor applications face significant technical challenges due to the invalidity of global position system(GPS)signal,imprecise localization,as well as the limited load,computing ability and flight endurance due to the mini size and possible on-board fuel/battery carried by the MAV.Therefore,this thesis focuses on the indoor application of MAV and deals with the problems of the realization of three-dimensional(3-D)panoramic environment reconstruction,high-precision localization and energy-optimal path planning without relying on any external navigation aid.The main research work and contributions of this thesis are as follows.(1)A 3-D panoramic environment reconstruction approach based on visual sensor network(VSN)is proposed.First,a 3-D VSN system composed of multiple RGB-D sensors is deployed in GPS-denied indoor environments.To fill the missing information on the original depth images,a black hole filling algorithm based on background depth image and k-neighborhood color matching method is present.Then,using only RGB-D data from the Microsoft Kinect sensors as input,the improved speeded up robust feature(SURF)extraction algorithm,the random sample consensus(RANSAC)estimation algorithm and iterative closest point(ICP)based 3-D reconstruction algorithm are utilized to build a fast 3-D panoramic model of the observed environment with improved accuracy.(2)A multi-sensor data fusion localization approach based on distributed estimation algorithm is proposed to improve the accuracy of the global localization.First,a target detection method based on foreground color and image segmentation is present to recognize the MAV target from the background.Then a distributed estimation algorithm known as kalman-consensus filter(KCF)is used to estimate the global position and the trajectory of MAV by fusing the estimations of all sensors that capture the target and coming to a consensus about the more accurate global optimal localization of the MAV.(3)On the basis of analysis and actual measurement results of the MAV energy consumption,a realistic,reasonable and accurate energy consumption model of a specific quadrotor is built to describe the relationship between the energy consumption and specific flight speed and motions.First,the energy consumption of each module of the specific quadrotor Crazyflie is analyzed.Then according to the force analysis and kinetic energy law of the quadrotor in the uniform linear flight,the function of the energy consumption varies with flight speed is established.The most energy-efficient speed can be obtained by computing the partial derivative of the function.Finally,the energy consumption model is established by characterizing the path by path length,climbing/descending rate and turning angle.(4)An energy-optimal 3-D path planning approach that minimizes the energy consumption while also considering the motion uncertainty is proposed.Instead of correcting motion deviations continuously,we exploit the trend of movement drift and a two-dimensional(2-D)Gaussian distribution function is employed to simulate the random drift and position bias.A modified heuristic ant colony optimization(ACO)is utilized to search the energy-optimal path traversing the 3-D map based on the energy consumption model and the Gaussian distribution.In the end,both software simulations and practical indoor flight experiments are conducted to validate the feasibility and effectiveness of the proposed approaches.The results demonstrate that our approaches achieve a fast,reliable and comprehensive 3-D map of the indoor environment,higher accuracy of localization and the planned path is more optimized in terms of the energy consumption.The proposed approach of this thesis is of great scientific significance and also has a wide range of practical application prospects in industrial sectors,everyday life,security monitoring and other fields.