A contribution to collision free trajectory planning for 6r robots in varying environments

Abstract

In flexible und complex handling systems trajectory planning becomes increasingly demanding. As an example, the geometry of the work pieces tends to vary more than it used to. Therefore, the planned trajectory of the handling robot has to either be adjusted for each piece individually or the trajectory has to allow general geometry. This leads to performance loss for smaller work pieces since they are less sensitive to obstacles and can therefore be moved on shorter (faster) trajectories. In both cases, time and costs could be reduced. Another characteristic of handling systems is that several handling robots share the same workspace. Again the positions of these obstacles are only known at runtime. This places great demands on the planning tools. In this paper, it is shown that both problems can be treated with a hybrid method called curve flow method. This method combines a gradient flow which comes from a variational principle and a potential field method. The variational principle makes the planned trajectories short, the potential field prevents collision.