Research On Three-dimensional Force Decoupling And Sliding Detection Method Based On Distributed Tactile Sensor

With the advancement of science and technology and the increasing demand for intelligent life,intelligent bionics and humanoid robots,which can realize humancomputer interaction,have also developed unprecedentedly.The tactile sensor can imitate the human tactile perception ability to a certain extent,and extract all kinds of information in the contact process,but this information does not directly improve the working ability of the bionic or humanoid robot.How to treat the unprocessed obtained from the tactile sensor The initial data,converted into useful high-level information,is one of the key steps to realize the intelligent robot's normal human-machine work and human-computer interaction.The flexible three-dimensional force decoupling of the flexible tactile sensor is very important in realizing the sliding detection of the robot and the intelligent grasping.This thesis summarizes the previous experience,based on the previous research,for the current problem that the tactile sensor cannot have both flexibility and threedimensional force detection,relying on the Chongqing Academy of Chinese Academy of Sciences designed a new type of distributed carbon nanowall flexible tactile sensor,The material structure of the sensor and the design scheme of the flexible FPC are introduced.The performance test of the sensor is carried out to ensure the reliability of the sensor device.And a special HTM sensor data acquisition hardware system based on STM32 is designed.Finally,the principle and process of data acquisition are introduced.For different three-dimensional force decoupling algorithms,This thesis introduces the threedimensional force decoupling process of BP neural network and BAS-BP neural network in detail,and uses RBF and GA-BP algorithms to make relevant comparison experiments.The final draft is selected.The optimal three-dimensional force decoupling algorithm is presented.Finally,using the relevant equipment and conclusions of the second and third parts,the sliding detection experiment of the robot during the gripping process is added.Using discrete wavelet analysis,the corresponding sliding threshold is obtained.Using the set sliding threshold can steadily determine whether sliding occurs during the gripping process,which is convenient for operations such as increasing gripping force.The results show that the new distributed carbon nanowall tactile sensor has the advantages of flexibility,high sensitivity,and large range compared with conventional tactile sensors.Compared with conventional algorithms such as BP neural network,the BP neural network algorithm(BAS-BP)based on the optimization of the beetle whisker search algorithm can effectively reduce the error of nonlinear fitting.During the grasping process of the manipulator,discrete wavelet analysis is used to analyze the decoupled three-dimensional forces,and the corresponding sliding threshold can be obtained.The set threshold can be used to accurately determine whether the object slides during the grasping process.Intelligent crawling laid the foundation...