Selfoptimized assembly planning for a ROS based robot cell

Abstract

In this paper, we present a hybrid approach to automatic assembly planning, where all computational intensive tasks are executed once prior to the actual assembly by an Offline Planner component. The result serves as basis of decision-making for the Online Planner component, which adapts planning to the actual situation and unforeseen events. Due to the separation into offline and online planner, this approach allows for detailed planning as well as fast computation during the assembly, therefore enabling appropriate assembly duration even in nondeterministic environments. We present simulation results of the planner and detail the resulting planner's behavior. © Springer-Verlag Berlin Heidelberg 2012.