Research On Obstacle Detection And Obstacle Avoidance Mechanism Based On Deptt Camera

With the development of information technology and computers,artificial intelligence has gradually become a research hotspot in various fields.As a typical representative of artificial intelligence,mobile robots have entered a new research boom.Detection and avoidance of obstacles are basic functions of the robot during movement.At present,there are many problems in obstacle detection when using ultras onic,infrared,laser radar,color camera and multi-sensor fusion,such as the inab ility to obtain boundary information of obstacle target,the vulnerability to light,the inability to obtain distance information,the large amount of calculation,the complexity of fusion algorithm,and high requirement on sensor synchronization.Based on the current problem,a method of using a depth,based on the current problem,a method of using a depth camera as a sensor to detect obstacles is proposed.The two-wheeled mobile robot Robot-BX is used as a research platform to detect obstacles.The problems in the obstacle avoidance have been studied in depth,and the corresponding solutions are proposed.The main contents of the thesis are as follows:Firstly,the obstacle avoidance method of the depth camera sensor is studied.Due to the large number of pixels in the depth image detected by the depth camera,the data processing efficiency is low and the detection speed is slow,especially not suitable for detecting moving obstacles.This thesis proposes a method to improve the detection speed of the depth camera.To screen the pixel points of the depth image in order to reduce the pixel points,and improve the data processing efficiency.In addition,due to the influence of background and ground,depth camera fails to accurately identify the specific location of obstacles.Meanshift algorithm is adopted in this thesis for image segmentation and background removal,and then obstacles are extracted by region growth algorithm.Secondly,the obstacle avoidance mechanism of the robot is studied.This paper proposes a method of detecting obstacles through the depth camera during the walking process of the robot,and then transmitting the information to the lower position machine to guide the robot to avoid the obstacles;for the single genetic algorithm and the ant colony algorithm,the obstacle avoidance encounters the dynamic obstacle avoidance The problem is that the speed of the obstacle is slow and the point of the inflection point is close to the obstacle so that it will collide with the obstacle.A new obstacle avoidance method is proposed,which is an obstacle avoidance algorithm combined with genetic algorithm and ant colony algorithm.The bottom layer of the robot uses the genetic algorithm to perform global search,and generates the global descendants,and transmits the generated particles to the upper layer.The ant colony algorithm's strong local search ability is used to improve the obstacle avoidance speed and the obstacle avoidance performance at the inflection point in the dynamic environment.Finally,on the Robot-BX of the self-built mobile Robot,depth camera sensor is selected to carry out the actual operation of the obstacle detection algorithm proposed above and the improved hierarchical ant inheritance-ant colony algorithm,and the feasibility of the experimental scheme is verified through the concrete object.The experimental results show that the method and mechanism proposed in this thesis are feasible.