Model-Based Testing of an Industrial Multi-Robot Navigation System

Nowadays multi-robot system getting more and more deployed in industrial settings. The usage of such a system is motivated by the fact that it allows flexible solutions which scale well with the current work load. One example of such a multi-robot system is a fleet of transport robots in a warehouse. The use of planning and scheduling algorithm allow such a fleet to maximize the throughput in the warehouse with a minimum of interference. To maximize the utility of such systems they have to operate on a 24/7 basis. Therefore, these systems need to show a high degree of dependability. The multi-robot system we are addressing in this work is motivated by a real industrial use case of a fleet of transport robots in a warehouse. The robots execute transportation tasks which are assigned to them in an automated way. To perform this transportation task without interference within the fleet and to respect operational restrictions of the environment a hierarchical planning approach is used [5]. The approach uses traffic areas which are defined by the user to model the allowed behavior of a robot within a particular area of the environment. This is done by treating areas like shared resources with different options for their utilization. A proper functioning of the navigation component is crucial for the dependability of the complete robotic fleet. Thus to ensure the high degree of dependability one needs to take proper measurements during the complete life cycle of this component. One way which allows to treat all phases of the development in a unified way and offers the possibility of a high degree of automation is model-driven engineering [2]. To apply the idea of model-driven engineering to autonomous robots one can follow the idea outlined in [7]. The authors propose to use model-based techniques during the life cycle of the robot ranging from model-based testing to model-based supervision. With the help of model-based techniques tedious tasks such as testing and supervision can