Research On Motion Planning And Collision Avoidance Control Method Of Mobile Robot Arm Aided With Recurrent Neural Network

Due to the high flexibility and large operating space of the mobile robot arm,it has been widely used in practical engineering fields such as welding,painting and assembling.In the actual assembly of the mobile robot arm,to reduce energy consumption and improve productivity,the position and end position of the mobile robot arm is needed to coincide with the initial shape as accurately as possible,i.e.,the mobile robot arm completes repetitive motion tasks.In addition,the mobile robot arm will encounter obstacles when performing complex tasks,which are required to complete the trajectory tracking and cooperative control and avoided collision with obstacles.Moreover,in actual working conditions,there are electromagnetic interference,model interference and various noises in the signal transmission process that commonly interfere with the movement of mobile robot arms.These perturbations can have an impact on model solving accuracy,stability and task completion,resulting in repetitive motion and collision avoidance task failure of the mobile robot arm.As a result,it is essential to design a non-convex noise-resistant recurrent neural network algorithm to solve the problem of repetitive motion,trajectory tracking and obstacle avoidance control of a mobile robot arm with perturbations.The main contents of this paper,which is an in-depth study of the above problems,are as follows:(1)For the motion planning problem of the mobile robot arm,the kinematic equations of the mobile robot arm based on the world coordinates are obtained by using the D-H method and the spatial coordinate transformation matrix,which lays the foundation for the subsequent solution of the motion planning of the mobile robot arm.(2)The non-convex velocity constraints,trajectory tracking and orthogonal repetitive motion of the mobile robot arm are transformed into a time-varying quadratic programming(TVQP)problem by means of quadratic performance indicators,physical constraints and kinematic equations of the mobile robotic arm.A recurrent neural network is used to solve the TVQP and control the mobile robot arm to complete the repetitive motion task.The orthogonal repetitive motion TVQP method can eliminate position errors and improve control accuracy.Numerical simulations show that the recurrent neural network model can control the mobile robot to complete the repetitive motion task accurately.(3)Repetitive motion,velocity constraint,and trajectory tracking of the mobile robot arm are transformed into an TVQP problem by means of quadratic performance indicators,velocity constraints and kinematic equations of the mobile robotic arm.The non-convex constrained noise-resistant zeroing neural network(NCNSZNN)is utilized to solve the TVQP problem and control the mobile robot arm to accomplish the repetitive motion,velocity constraint,and trajectory tracking tasks in the presence of external disturbances.The theory demonstrates that the NCNSZNN model can suppress external noise perturbations.Numerical simulations and platform experiments verify that the NCNSZNN model can eliminate external noise disturbances and enable the mobile robot arm to accomplish trajectory tracking,velocity constraint and repetitive motion tasks.(4)Repetitive motion,trajectory tracking,physical constraints and collision avoidance are transformed into an TVQP by means of quadratic performance indicators,physical constraints and kinematic equations of the mobile robotic arm.The primal dual neural network is exploited to solve the TVQP problem and control the mobile arm to accomplish collision avoidance,trajectory tracking,joint constraint and joint velocity constraint.Numerical results and platform experiments verify that the primal dual neural network model can control the mobile robot arm to accomplish trajectory tracking,repetitive motion,joint constraint,joint velocity constraint,and collision avoidance tasks.