A Fast Search Method for Computing the Constant-Orientation Workspace Based on the Binary Approximating Principle

Abstract

In order to compute the constant-orientation workspace of serial robots, a fast search method based on the binary approximating principle is proposed. On the basis of binary tests of the inverse kinematics and the collision detection of a robot, this method introduces virtual straight lines containing the orientation information and selects two points(one locates inside the constant-orientation workspace, while the other outside) to carry out the dichotomy and test processes, so as to approach the boundary of the constant-orientation workspace step by step. Firstly, the concept and the state of the art of constant-orientation workspace are introduced in brief. The constant-orientation workspace problem is abstracted, followed by the basic solving ideas. Then the binary approximating search method and its applications to computing the 1D、2D and 3D constant-orientation workspace are presented in detail. Finally, the effectiveness of the proposed method is demonstrated by simulations and comparisons, showing that the method achieves the accuracy as 1 mm consuming about 14.5 ms to search the 1D constant-orientation workspace, and is suitable for solving the 1D ~ 3D collision-free constant-orientation workspace problem of robots with different degrees of freedom or configurations.