Research On Autonomous Obstacle Avoidance Algorithm For UAV Based On Moncular Vision

In the civilian or military field,uav has been widely used,and uav want to continue the development,in addition to endurance functional progress,tne safety aspect is not to be underestimated,so autonomous obstacle avoidance of uav has become a hotspot about uav research in recent years and uav have autonomous obstacle avoidance capability is also realized the premise of the uav autonomous navigation movement. no matter in theory and in practice uav autonomous obstacle avoidance technology have achieved fruitful results?but its application sensors such as some heavy and expensive sensor like laser radar and binocular vision. So in recent years,researchers have tried to use a single camera as the primary sensor,that is to say moncular vision method.Mainly because of the small uav load caoacity is limited,and the use of onboard cameras than the use of other sensors(such as laser radar and infrared sensors) can bring additional benefits,it can reduce the overall computing time and battery consumption,and it is possible to achieve real-time vision-based obstacle avoidance algorithm.First,the research status of uav autonomous obstacle algorithm are introduced,it pointed out the advantages and disadvantages of the prior art,then combined this study,and introduce multi-scale invariant characteristics of the image.This study have conducted a detailed theoretical analysis to the current two popular features operator(SIFT and SURF operator),and conducted comparative experiments.Next,this study introduce the coordinate system eatablish and attitude angles in the definition of the four -rotor uav used in the experimental,and the main hardware onboard inertial measurement device constitutes has a brief understanding.The theory of how to use the imu data combine with complementary filtering algorithm to solver current posture ,and uav attitude control method were outlined,and have actual attitude solver algorithm related experiments based on the obtained data.Camera imaging principle is analyzed,and a bried introduction of the camera imaging principle,and the camera model of linear and nonlinear are described,and the camera calibration tool was used to calibrated so that the inner parameter matrix for the algorithm behind use.This study proposed amodel to calculate the distance between the obstacle to uav,accorading to this distance and put forward the concept of a security zone,when the obstacle into the secure area of the aircraft,control signal changes the uav attitude for obstacle avoidance .Proposing two obstacle avoidance methods are vertical and lateral obstacle avoidance strategy.Finally,the uav related indoor static and dynamic ranging preliminary experiments verify the correctness of ranging model.Using the off-line methods confirmed the feasibility of obstacle avoidance strategy.