Research On Obstacle Avoidance Strategy Of Mobile Robot Based On RGBD Image

With the development of economy and the gradual increase of urban population,nurses in urban central hospitals have become more and more arduous in clinical nursing,which cause the contradiction between nurses and patients.In order to liberate the nurses from the usual tedious work and provide better services for patients,we must take the nurse assistant robot into hospital,aimed to help nurses to work convey goods.The nurse assistant mobile robot which must completes the work of transporting goods safely and reliably faced extremely challenge.Therefore,this paper studies the problems of pedestrian detection,object tracking and obstacle avoidance in indoor dynamic environment,which develops and verifies effectively based on ROS system.The obstacle avoidance strategy process: first,using human detection algorithm detect the region of people;second,tracking this region that we detect;last,we can get the position of obstacles and determine the motion state of targets,by which we can complete the obstacle avoidance behavior of local path planning based on improved artificial potential field method.Specific work is as follows:The proposed algorithm process was as follows: first,efficient clustered analysis to get the set of point cloud based on the density of information;then according to the height of human existence we divided into standard area in which the people may exist;finally based on the SVM-HOG framework for human detection.Experiments show that the above scheme can reduce the error detection rate and the computation of the computer so as to achieve real-time requirements.Online Boosting algorithm is used to transform the target tracking problem into a two value classification problem.In this algorithm,the Haar-like classifier is trained and updated to obtained the needed classifier in real-time.When the present detection sample arrives,the classifier is used to discriminate the sample and repeat the previous step to update in real time.In addition,greedy data association method is introduced to solve the problem of how to re track the target which is lost.Based on the interactive multiple model(IMM)filtering algorithm,the motion state of obstacles is estimated accurately,and the local dynamic path planning is performed by using the improved artificial potential field method.In consideration of estimate the obstacle position inaccuracy by using Kalman filter,we use a new filtering algorithm based on interacting multiple models is proposed.We use the improved artificial potential field method to avoid obstacles,which considerating the speed of people,compared with the traditional method,the proposed algorithm can better finish the robot obstacle avoidance in a dynamic environment and reduce the rate of collisionwith obstacles.Comprehensive experiments show that the nurse assistant robot designed based on the above strategy can deal with the indoor dynamic obstacle avoidance under the condition of the normal walking speed of the sparse crowd.Compared with the traditional obstacle avoidance strategy,the robot has higher planning efficiency and more intelligent and safer handling of pedestrians.