Research On Cooperative Control Of Manipulator Driven By Dynamic And Static Gesture

With the expansion of the application scope of industrial robot/manipulator,the manipulator operation task is more and more complex.The traditional human-machine interaction mode such as joystick and operation panel has been unable to meet the robot’s various needs of building tasks.Therefore,in the manipulate control of the manipulator,gestures as an intuitive human-computer interaction have received extensive attention and research in recent years.In the application of hand gesture in manipulator control,The gesture led absolute instruction makes it difficult to complete the complex task;and the lack of systematic description of dynamic gestures and static gestures makes it difficult for the manipulator to understand the complete intention expressed by the operator.In view of the above problems,this paper proposes a dynamic and static gesture driven sharing control method for robot manipulators,so as to give full play to the intelligent decision making ability and autonomous perception planning ability of robots in task execution.The main contents of this paper are as follows:A dynamic and static gesture recognition method based on deep image is studied.First,the depth image is obtained and preprocessed based on the Kinect.Based on the Hu invariant moments feature extraction of static gesture contour,the deep neural perception network model is used to improve the classifier model,and the static gesture recognition model based on single frame depth image is constructed,which improves the recognition rate of static gesture.Aiming at the dynamic gesture angle centroid sequence recognition,gesture recognition led to the speed of movement of the centroid accuracy is not high,the limited distance sampling method based on centroid sequence,and constructs the dynamic gesture recognition model based on Hidden Markov model,guarantee the dynamic gesture recognition accuracy and robustness.The construction method of hand driven robot hand shared control task instruction is studied.First,a shared manipulator control model based on dynamic and static gesture recognition is proposed to fully play the human intelligence decision making and robot autonomy in the process of task building.Then the multi-level manipulator shared control instruction set is designed and the compiler is implemented.Finally,gesture and instruction to instruction mapping,the manipulator built to describe task,complete the self-generated gesture driven manipulators sharing control,decoupling from the task of building the task execution of this process,as the task execution reserved space optimization.The multi-level autonomous analytical execution method for the shared control task of the manipulator is studied.First,the kinematics modeling of the manipulator is carried out by task analysis.Then,based on the task syntax tree,the robot’s shared control task is analyzed autonomously at multiple levels.First,we plan the job order based on the task level knot tree,and then generate the multipath path of robot manipulator based on the analysis of action level node.Finally,give full play to the manipulator underlying autonomous perception planning,aiming to guarantee task safety,the motion path vector of the manipulator is optimized and adjusted based on the virtual potential field perception.Aiming at the rigid impact problem caused by multi-path stop-acceleration transition,based on circular interpolation and mixed multi-path,reduce joint stop-acceleration times to reduce rigid impact.A manipulator shared control system driven by hand gestures is implemented and verified.This paper studies the static and dynamic hand gesture recognition,manipulator shared control task instruction construction driven by gesture method,the manipulator control method performs the task sharing multi-level analysis based on independent,starting from the functional requirements of the system analysis,the overall framework and the man-machine interface design of the system,from the simulation and control entity manipulator carries out two aspects of the task verification of shared control system.