Research On Demonstration Programming Method Of Multi-joint Robot Based On EMG Signal

In recent years,the research of intelligent robot technology has become more and more indepth.Demonstration programming has become another hot topic in research.Demonstration is a method of learning human and animal skills.It gives the demonstration learning mechanism to the robot system to make it similar.The motor skills are one of the important aspects of demonstration programming research.Industrial manufacturing is a complex and necessary task,and it is not easy to use industrial robots to accomplish this task.In order to solve the above problems,this thesis has conducted in-depth research,the main research results are as follows:Firstly,in the aspect of acquisition and recognition of EMG signals,the neural mechanism of sEMG generation is analyzed.The IIR re-line filtering is studied,and then the time domain features of signals are extracted by different time domain methods.The comparison and comparison of the recognition results and the actual verification performance of the three classifications were carried out.Finally,the improved SVM and HMM secondary model classification algorithm was used to control the laboratory youbot robot,and the recognition rate reached 93.6%.In the process of controlling your omni-directional movement,the actual requirements are met.Secondly,in the aspect of human-machine coordinated control based on sEMG signal,this thesis firstly establishes the kinematics model of the laboratory Yumi robot,and analyzes the forward and inverse kinematics,and then analyzes the working space of the dual-arm robot.The motion planning analysis of the Yumi seven-degree-of-freedom manipulator is carried out under matlab using the fifth-order polynomial interpolation method.The error analysis of the end point and the actual arrival point is carried out,and then the model of the Yumi robot is established based on the ROS environment.Further verify the effectiveness of the algorithm.After that,the mapping of man-machine space is carried out to establish the positive kinematics of the arm.The seven-degree-of-freedom manipulator automatically plans to the position of the person's demonstration through inverse kinematics to realize man-machine coordinated control.The experiment was carried out using the laboratory's Yumi seven-degree-of-freedom robot,and the Yumi robotic arm movement was controlled in real time by wearing a bracelet.Finally,this thesis adopts the concept of Programming by Demonstration(PBD),and proposes a PBD framework to demonstrate the reproduction problem.Guided by this framework,PBD is divided into two parts:motion pattern recognition and robot control mode reproduction.The recognition is mainly based on the characteristics of the EMG information during the movement of the presenter.The above-mentioned two-level model classification algorithm is used to identify and classify the demonstrative action patterns.The robot control mode reproduction is divided into the analytical plan of the manipulator motion trajectory and the human-machine heterogeneous joint coordinates.The mapping and motion trajectory analysis planning is mainly based on the trajectory recognition algorithm(GMTI)of the proposed Gaussian mixture model to obtain the motion trajectory of the demonstrator mapped to the manipulator,and then the robot arm to the presenter according to the robot arm trajectory planning algorithm.The optimized reproduction of the demonstration trajectory,according to this framework,is verified on the laboratory Yumi robot,and the demonstration programming of the multi-joint Yumi robot is realized by demonstrating different motion modes.