Modeling and Simulation-based Development of Autonomy Features for Drones

摘要

In the last decade, more and more aerial robotics researchers show interests in developing autonomy features for drones to solve problems in different areas. But the development of autonomy features is complex and labor intensive. Accordingly, model-based design and simulation-based verification is becoming an industry standard in development of autonomous airborne systems. This we call modelling and simulationbased development. However, commercial model-based design and simulation tools and supporting testing environments require a considerable amount of investment. In oder to provide a more economic and efficient solution, this paper investigates a pipeline for modeling and simulation-based development of autonomy features for drones using open source software and hardware stacks. In this context, a generic drone architecture is being designed based on open source hardware platforms, namely CC3D and Raspberry Pi. In the software stack, LibrePilot, an open source software suite to control multicopters is extended to support the designed architecture. The design of the autonomy features is developed using the model-based design in Scilab/Xcos. Xcos Reuseable and Customizable Code Generator is utilized for automatic code generation. The software stack will also include a generic plant model. The workflow starts from autonomy feature modeling and ends with flight testing through Model-in-the-Loop (MiL) testing, Software-in-the-Loop (SiL) testing, target deployment, Hardware-in-the-Loop (HiL) testing. The approach is demonstrated with a simple case study about an autonomous landing feature.