Control of robots handling an object in real time (French text).

摘要

This report presents an algorithm of position control for a 5-dof industrial robot grasping an object. This algorithm, developped by Slotine and Li, is employed to produce an adaptive position controller during the movements without constraints. This control law is made up from feed-forward and a feedback to a proportional derivative regulator (PD). The feed-forward is built starting from the model of the inverse dynamics of the robot. And the PD controller is established by sensing articular speeds and accelerations. Moreover, a position controller of proportional derivative was implemented in real time on QNX platform. The parameters of the controller were adjusted to minimize the tracking errors. The implantation in real time is based on the macro-functions of the S626 I/O card of Sensoray for position sensors and the PCIDCC5-P card of ICS for sending the PWM signals. Also, an articular trajectories generator, starting from the rectilinear cartesian ways, was developed using the Bounded Deviation Method (BDM) of Taylor and by exploiting the most recent tools including the techniques of the Banded Diagonal Matrix ( BDM). The simulation results of the adaptive control applied to the US MAKER 100 robot, which show the stability and the good performance of the proposed controller, also illustrated in a graphical environment developed within the framework of this project. This interface helps the user to define the desired trajectories in cartesian space, and to modify the control parameters and to visualize the movement.