Research On Target Searching And Positioning Control For Multi-rotor Aircraft

The multi-rotor aircraft integrates the achievements of Micro-electro Mechanical Systems(MEMS),computer technologies,control theories and artificial intelligence etc.In recent decades,it received more and more attentions.With the enhancement of high autonomy,it was applied widely in military and civil fields with good application prospect,including search and rescue,border patrol,aerial mapping,target tracking,communications for disaster recovery,power network monitoring,traffic detection and air pollution detection etc.Although the autonomy increases progressively,multi-rotor aircraft still has its bottle-neck.For instance,in the application of target searching and positioning,autonomous navigation could not identify obstacles immediately under the premise of no pre-planning and make instant decisions at the intersections.Generally,people use hands to control the aircraft,which led them unable to accomplish more control tasks simultaneously.Additionally,for the unpredictable complexity in practical application,human interference is sometimes necessary.However,the different skill levels of people affect the aircraft control differently.Therefore,it is a problem to be further investigated how to introduce the more intelligence,stability and easy controllability modes in target searching and positioning of the multi-rotor aircraft.To realize the target searching and positioning,this thesis performed systematic and in-depth research on multi-rotor aircrafts.The optical flow was utilized to realize the multi-rotor aircraft control for hovering,vertical landing and terrain tracking firstly.Next,the indoor and outdoor target precise positioning was achieved by systematically studying on Strap-down Inertial Navigation System(SINS),Global Positioning System(GPS)and visual navigation system.Then,the Brain-Computer Interface(BCI)system was proposed to implement two-dimensional(2D)space target searching using the semi-autonomous navigation system and Electroencephalogram(EEG)control.Finally,the Hybrid Computer Interface(HCI)system was presented to achieve 3D space target searching based on BCI system using the EEG control and Electrooculography(EOG)control.This study was mainly developed and researched in the following five aspects:1.Research on multi-rotor aircraft control based on optical flowAccording to the dynamic equations,three different nonlinear controllers were designed for hovering,vertical landing and terrain tracking.The stabilities of nonlinear controllers were analyzed and proved using Lyapunov method.The optical flow of target points was obtained by comparing the two time continuous images using the pyramid Lucas-Kanade algorithm and utilized as the feedback.This thesis also presented an auxiliary hovering control system by monitoring aircraft’s position,attitude and velocity information to eliminate the hover failure(deviations on unknown directions of aircraft in horizontal plane)caused by special environments(e.g.aircraft unable to get optical flow).The experimental results showed that compared to other similar methods,the methods proposed in this thesis in hovering,vertical landing and terrain tracking are more accurate in location precision.The obtained optical flow can be controlled varying at the vicinity and converging to expectations.2.Research on multi-rotor aircraft positioning based on integrated navigation system with SINS and double GPSTo achieve the multi-rotor aircraft outdoor precise positioning,the integrated navigation system based on SINS/double GPS was researched and utilized to design the online estimation system and methodology of complex terrain boundary and area.The proposed online estimation system and method overcomes the high cost,low estimation precision,low efficiency and landscape limitation drawbacks in other methods.It could be applied to any terrain without intersection points.The SINS and double GPS relative pseudorange differential positioning systems were constituted the integrated navigation system and aimed at obtaining the positioning information.During getting positioning information,abnormal data may be produced from aircraft unstable hovering.Therefore,Pauta principle was used to rule out the outlier data firstly.Then,the Extended Kalman Particle Filtering(EKPF)was used to increase the positioning accuracy of border points.The true experiment indicated that the positioning accuracy of border points reaches sub-meter level by using EKPF and the area estimation accuracy is within ±1%.It increases the accuracy by 1-3.6%compared with other similar methods.Furthermore,it verified that the algorithm is feasible and effective.3.Research on multi-rotor aircraft positioning based on integrated navigation system with landmark-based visual navigation and SINSTo overcome the drawbacks of GPS,the integrated navigation system based on landmark-based visual navigation and SINS was researched and employed to realize the multi-rotor aircraft indoor precise positioning and path planning.The visual navigation was achieved by matching the known geographic image database and the current image obtained by aircraft.Firstly,the image matching process was accomplished through the Scale-invariant Feature Transform(SIFT)which is used to fulfill point feature extraction.Then,the methods of Random Sample Consensus(RANSAC)and Feature Group Matching(FGM)were applied to complete filtering.To avoid obtaining the positioning information only depending on SINS for a long time,the navigation performance of aircraft was considered in the path planning algorithm and at least one landmark should be entered the vision field of camera at any time.Moreover,the Dijkstra algorithm is extended and adopted to calculate the shortest path network between multiple start points and destination points.The experiments indicated that the resulting path of aircraft flew along the obtained final shortest navigable path is within the permissible flying area without any landmark missed.This proves that the proposed algorithm is applicable to the indoor precise positioning and path planning of multi-rotor aircraft.4.Research on 2D space target searching based on semi-autonomous navigation and EEG controlTo overcome the defects of autonomous navigation system on target,to make people able to accomplish more control tasks simultaneously and introduce the more intelligence,stability and easy controllability modes,the BCI system based on semi-autonomous navigation and EEG control was proposed.This semi-autonomous navigation subsystem was consisted of a 2D laser range finder and a front facing CMOS camera with 90° angle lens.It was used to provide feasible flying directions for decision subsystem which was constructed on the basis of EEG feature analysis of left-and right-hand Motor Imagery(MI)tasks.Simultaneously,it was utilized to realize multi-rotor aircraft semi-autonomous obstacle avoidance according to the feasible flying directions.The resulting BCI system used Independent Component Analysis(ICA)to isolate and to remove the artifacts from the collected six channels EEG signals.Additionally,it used the improved Cross-correlation(CC)method to accomplish the MI feature extraction and utilized the Logistic Regression(LR)method to complete the MI feature classification and decision.The practical target searching experiment indicates that subjects can master the MI tasks in shorter learning cycle.It proves that the BCI system has good adaptability.Compared with mobile phone control,the flying path of BCI system is more smooth and regular.It proves that the BCI system has better control stability.Concurrently,it verifies that using the BCI system to achieve the multi-rotor aircraft 2D space target searching is feasible and effective.5.Research on 3D space target searching based on EEG control and EOG controlTo solve the problem of searching space restriction,the HCI system was proposed to realize 3D space target searching using the EEG control and EOG control based on BCI system.It used the 1D wavelet decomposition at level ten to remove the eyeblink EOG baseline drift and applied the median filtering method to eliminate EOG noise.The Continuous Wavelet Transform(CWT)method was employed to indentify eyeblink features which were used to switch interfaces between horizontal and vertical MI tasks.The HCI system utilized ICA method to kick out artifacts from the collected six channels EEG signals.The HCI system employed the Joint-regression(JR)model and spectral powers method to extract time and frequency domains features of MI tasks.Simultaneously,the Support Vector Machine(SVM)was employed to complete the MI tasks features classification and final decision.The practical target searching experiment indicates that subjects can master the related eyeblink and MI tasks in a shorter time.It proves that the HCI system has good adaptability.Compared with mobile phone control,the flying path of HCI system is more smooth and regular.It proves that the HCI system has the better control stability.Concurrently,it verifies that using the HCI system to achieve the multi-rotor aircraft 3D space target searching is feasible and effective.The researches on multi-rotor aircraft target searching,positioning and control in this thesis are contribute to solving the defects of autonomous navigation system on target searching and providing a novel solving plans for target searching and positioning.The methods in this thesis are of value as a reference in both theoretical and practical applications.