Research On The Family Daily Tool And Indoor Area Classification Algorithm For Functional Cognition

With the development of robot hardware and the rise of artificial intelligence technology and the increasingly serious social aging problem,robot into the family for the human services,has become an inevitable trend of social development.Everyday activities of people will involve a variety of tools,it's very easy for human beings to interact with their familiar environment,but still is a research hotspot and difficulty in the field of intelligent robot development.This paper studies the function cognition of the robot,which aims to give the robot tool cognition and the ability to recognize the environment.First of all,building robust underlying features is the foundation of image research.Fusion of different features will produce redundant information while enhancing image description.Aiming at the above problems,this paper proposes a feature optimization algorithm based on many geometric features.Based on the feature combination,each functional component of the tool is the description object.The feature selection model of the ReliefF algorithm is the framework,and the best feature description combination of the components is selected.Compared with the traditional feature selection,this method realizes the combination of different feature extraction methods.Secondly,the best description of the functional components is obtained by the feature selection,which is the basis for the effective description of the tools.Aiming at the problem that robots still cannot realize the overall cognition of tools,this paper proposes a kind of tool classification model which is fast and reasonable for the hardware configuration of the robot.On the basis of feature optimization,based on the bag of word model to construct the bottom-up cognitive framework of the tool.Compared with the traditional research on tool classification,this paper is no longer limited to the superficial features of tools,but a definition method based on functional components.Finally,the household daily tools will be distributed in different indoor functional areas,giving the robot the cognitive ability to the environment,which is an important prerequisite for the robot to find the necessary tools quickly and accurately.Aiming at this problem,this paper further proposes an indoor functional area classification model basedon CLM model.This model uses the image description method which avoids codebook,and directly constructs the Gaussian model to the underlying features,reduces the quantization error,and obtains the more stable description of the image.