کالیبراسیون قاب پایه برای ربات های صنعتی هماهنگ شده

To solve the problem of base frame calibration for coordinated multi-robot system, a new method is proposed in this paper. It is carried out through a series of “handclasp” manipulations between two coordinated robots, then a preliminary result can be reached by the calibrating equation. After that, in order to make sure that the calibrated rotation matrix is orthonormal, an optimal estimation of the relative rotation between the base frames of coordinated manipulators is solved out under the criterion of optimal Frobenius norm approximation. By the quaternion representation for rotation matrix and the Lagrange Multiplier method, an orthonormal matrix can be reached which is just the unknown calibrating result for base frames of the coordinated robots. Simulation and experiment results have verified the validity and effectiveness of the proposed method.

Great progress has been made after the robots were introduced into manufacturing industry. Nowadays, modern production needs more complex process which makes great challenge for traditional single robot system, so the multi-robot manufacturing system is on the way. At present, the multi-robot coordination system is one of the most challenging frontier in robotic research. In order to complete the coordinated control, position and orientation transformation relations between the base frames of the coordinated robots is required to satisfy the constraint relation of the robot end-effectors.

A simple but effective procedure for calibrating the relative rotation matrix and translation vector between base frames of two coordinated robots is presented. The greatest advantage of our approach is that there is no other measuring apparatus required for the calibration procedure, which makes it quite feasible for applications in manufacturing field. The calibration procedure is based on the robot joint information of a series of handclasp manipulations. Owing to driving the TCP point of each coordinated robot to one same point in the workspace for four times, the problem of determining the relative rotation matrix can be converted to solving a calibration equation of the form A=XBA=XB for the rotation matrix XX. After the rotation matrix was determined, the translation vector can also be solved out by point transformation equations. The only constraint for this procedure is the four points used here must be different with each other and noncoplanar, which is not difficult to satisfy for any coordinated system. In order to ensure the calibrated result satisfy the orthonormal constraint for rotation matrix, an orthonormalization procedure for the preliminarily calibrated result is also discussed. By using the quaternion representation for rotation matrix and solving a nonlinear equation with Levenberg–Marquardt method, a refined calibration result can be reached, which is also the optimal Frobenius norm approximation of the preliminarily calibrated result.