Design, Implementation, and Validation of a ROS2 UAV Simulator for Research Applications

This paper outlines the development and validation of a modular UAV dynamics simulator developed with ROS2, aimed at facilitating design analysis, control testing, and future integration of reinforcement learning for hybrid VTOL platforms. The simulator prioritizes accuracy and modularity with the goal to give researchers the ability to simulate aerodynamic forces, inertial dynamics, and propulsion system behaviour for different types of UAV configurations. To validate the simulator’s accuracy, a two-stage trim-centric methodology is used. First, the equilibrium trim states and control inputs are extracted for hover and forward flight and compared against a trusted MATLAB/Simulink implementation. Second, open-loop trajectories are simulated from these trim points, and their state evolution is compared using RMSE and maximum deviation metrics. Results demonstrate strong agreement, with trim RMSE values below 10^-5 and negligible drift in dynamic propagation under hover conditions. Notably, larger deviation in inertial X-position under forward flight is attributed to numerical sensitivity in time integration rather than physical modeling mismatch. The existing limitations comprise the lack of closed-loop validation and visual rendering, which will be rectified in subsequent efforts via Gazebo integration, PID-based trajectory tracking, and enhanced trimming functionalities for transition flight. The findings validate the simulator's precision and provide a dependable basis for subsequent autonomy and control research.