Application Of Multi-Sensor Fusion Information Technology Based On Robot

Multi-sensor, information fusion and traditional robot compose the intelligent robot. In the past study of the field of robot's intelligence, due to the neglect of comprehensive analysis of the multi-sensor, gives an adverse impact on intelligent system. The paper puts forward that, through the method of information fusion, two different types of sensors can be combined well to achieve the goal of right and effective recognition of object.Firstly, the current study of multi-sensor information fusion and intelligent robot are introduced, the principle, characteristics, system structure, methods are described, neural network applied on information fusion is discussed, the paper also put forward the key problems and study's direction.Secondly, according to the subject, an object recognition system based on the neural network of information fusion is designed, overall function of system, the selection of sensors, system's structure and workflow are described.Thirdly, the input information of neural network are discussed that how it is extracted. On one hand, vision sensor's image pretreatment, features selection and compute are studied in detail. On the other hand, tactile sensor's information extraction and output are put forward. And then, that information from two sensors are unified.Fourthly, a BP neural network recognizer is designed, parameters selection and algorithm are analyzed, the paper proposes a method that improves BP algorithm, data structure and treatment functions.Finally, based on the VC++6.0, the paper develops a information infusion neural network system, and the application steps on Lake Clean-up Robot are described in detail.