Analysis of design strategies for unmanned aerial vehicles using co-simulation

abstract

Designing critical embedded systems, like UAVs is not a trivial task because it brings the challenge of dealing with the uncertainty that is inherent to this type of systems, e.g., winds, GPS uncertainty, etc. Simulation and verification tools that provide a level of confidence can help design such systems and increase the safety of specified cyber-physical systems before deployment. This paper presents a framework for evaluating flight strategies of UAVs. Our framework is constructed by integrating, using high-level architecture, Ptolemy, a high level specification tool, and SITL/Ardupilot, a domain specific UAV simulator. It allows to evaluate flight strategies under the presence of uncertainty, such as winds, with a level of confidence by constructing a sufficiently large number of simulations. Its effectiveness is demonstrated by testing two different flight strategies in two scenarios under different wind intensities. We measure the flight quality providing quantitative information about the quality of the tested flight strategy, such as distance traveled, with a confidence of 95% and error of 8%.