Design and development of a software architecture for autonomous mobile manipulators in industrial environments

During the last decades, methods for intuitive task level robot programming have become a fundamental point of interest for industrial applications. A robot programming framework is needed to facilitate task-level programming of mobile manipulators, e.g. by providing the robot with a set of movement primitives and skills. Robot skills have already been used and tested successfully within the FP7 project TAPAS for this purpose, and are presently used in several new FP7 projects (e.g. CARLOS, STAMINA, ACAT). To meet the goal of programming mobile manipulators on-the-fly inside industries we need to bring together the actual skills and primitives with other aspects such as task planning, an extended knowledge integration framework and a control infrastructure for the skill execution. In this paper we want to discuss an architecture that organizes clearly the software processes on the different abstraction levels and propose a world model that ground abstract reasoning to concrete skills. Our aim is to develop a software framework to get towards the implementation of flexible and highly modular cognitive robot tasks.