Online Trajectory Generation of a 2 Link Robot in Presence of Obstacle

Abstract

Online coordination of multiple robots working on a single workstation requires special attention. In these applications it is generally necessary that the robot arm follow a desired path in workspace so that it does not crash with any obstacle or the other robots. In this paper a two link planar robot crossing a rectangular obstacle is considered. The proposed idea is to define the relationship between obstacle's dimensions and the required joints trajectories parameters which allow the robot to reach its destination in the presence of an obstacle. First, the desired path for robot avoiding the obstacle is defined using a fourth degree polynomial. Corresponding robot joints trajectories are defined using a sinusoidal function with four parameters. Next, Design of Experiments (DOE) technique is utilized. Three levels for width and length of the obstacle are used as input and a full factorial DOE with nine experiments is defined. Instead of using the inverse kinematics, Particle Swarm Optimization (PSO) algorithm is used to obtain parameters of the robot joint sinusoidal functions. A second degree regression is used to obtain the relationship between each of the four sinusoidal function parameters and the obstacle dimensions. The obtained regression equations allow online changes to the trajectory as obstacle dimensions change. Four case studies, different obstacle dimensions, are simulated using the two link robot. The results show that using the obtained relationships the robot reaches its desired destination, with high accuracy, while avoiding the obstacles.