| With the rapid development of technology and robotic technology,it has become an inevitable trend for robots to enter into homes.In our daily life,the functions of grasping and delivery of items are quite often,but it is difficult for the disabled and disabled people who are inconvenient in movement,for example the elderly,and they need a robot to help.This dissertation studies object identification,object location,and sound source localization based on the NAO robot platform.Generally,objects are recognized via color attribute.However,this method has been found to be susceptible due to illumination variations and interference with homochromatic backgrounds.Hence,a matching recognition algorithm based on color component statistics is proposed in this dissertation.First,all the regions of interest including the target object are found in the underlying image.Then the interference objects are eliminated in each region.Finally,the target object is segmented thoroughly.In the target location by the NAO robot,since the NAO robot's cameras have almost no public view area,and can only use one camera at the same time,the location principle of monocular vision is accordingly applied to locate objects.This dissertation derives the relationship between the height change of the head camera and the head pitch angle of the robot based on the hardware structure of the robot head.In this case,the monocular location algorithm for NAO robot is improved.Experiments show that the location accuracy can be within 1cm.In the study of sound source localization,in view of the poor location accuracy of the traditional ESPRIT algorithm in the low SNR environment,an improved RANESPRIT algorithm is proposed in this dissertation.The simulation results show that the improved algorithm has a certain improvement in location accuracy and robustness in comparison with the traditional ESPRIT algorithm.By integrated aforementioned techniques,the whole process of grasping and transmitting target objects is performed. |
PDF Full Text Request