Research On Robot Real-Time Writing Control

With the development of industrial technique,robots have been used in wide applications.In particular,the robots play key roles in the dangerous environments.Currently most of the robots are used with the defined control programs to perform some specific tasks.But insome complex applications,robots are required to perform non-defined tasks such as imitating human motion in real time.In such case it arises more challenges.This thesis focuses on the key problems when the robots performs non-defined tasks.It writes according to the human handwriting in real time.The detailed work includes:Firstly,human motion tracking techniques and control algorithms for the robot arm are discussed.And the advantages and disadvantages are compared.In addition,the skeletal node capture of the Kinect sensor and the depth image technique are introduced in detail.Secondly,robot handwriting is studied following the tracked motion with Kinect.A jitter problem exists in the data of the hand joint point captured by the Kinect sensor,which is serious for handwriting.In this thesis,the hand geometric moment fitted by the depth image is used instead of the single hand joint point in the proposed dynamic hand region extraction method.Experimental results show that the error in the X-axis and Y-axis directions can be reduced about five times.In addition,to combat the problems of the large gap existing in the collected track points which results that the mechanical arm writing is not smooth,and the existing interpolation algorithm has high delay,the N reference point real-time interpolation method is proposed based on the trajectory interpolation theory.The results show that the handwriting is rather smooth with the interpolation method.Finally,based on the Vicon system,the position of the nib is tracked in real time to control the writing of the robot arm.To combat the problem of the large deviation during the nib tracking,a real-time tracking method is propsed based on the joint rotation of the mechanical arm and the rigid body invariance theory.Experimental results show that the average trajectory error is reduced from 9.603 mm to 0.004 mm by the tip tracking method,and the trajectory is much smooth.Besides that,to combat the problems that the mechanical arm might vibrate during writing and the mechanical arm is abnormally stopped due to the blocked data,a data "receiving-execution" synchronization mechanism and the closed-loop second fault-tolerant check mechanism of the mechanical arm are designed based on the idea of the circular queue and the closed-loop control mechanism of the robot.Experimental results show these scheme can effectively eliminate the abnormal stop of the arm and effectively reduces the execution delay of the arm from 10.578 ms to 5.256 ms.