Robotic control of sliding object motion and orientation

Abstract

The models and control strategies presented allow repositioning an object that slides on a surface or between the jaws of a parallel-jaw gripper. Such techniques can save the need to pick up payloads for moving, or to deposit a gripped object for grasp modification. Precisely controlling a sliding object's position and orientation requires a model of its motion that recognizes both imposed motion and frictional forces. This paper presents an analysis of sliding-object motion that is based in part on Mason's quasistatic integral for frictional force (1985). Mason's model is extended to create linearized models of both straight-line and curvilinear sliding motion that permit deriving feedback control strategies based on purely proportional feedback of position errors, relative to reference trajectories. These controllers are very robust due in part to the simplicity of the model and of the physical plant, despite the nonlinearity of the true plant dynamics. Application of integral control is straightforward, in order to correct for modeling errors or disturbance forces, and control saturations are suggested to ensure stability even for large errors. Closed-loop control simulations confirm the effectiveness and robustness of these feedback control strategies.