In this chapter, we will describe in detail the steps required to set up the RotorS simulator framework, shown in the following screenshot, including the Robot Operating System (ROS) and Gazebo. Once this chapter has been completed, we will be able to set up the simulator, attach basic sensors to a MAV, and make it able to navigate autonomously in the virtual world:

We will also be able to compare the algorithms using the evaluation scripts. Finally, we will discuss how all the aspects learned and methods developed in this chapter could be then applied to a real MAV.

An overview of the primary components of the RotorS simulator is shown in the following diagram. Although, in this chapter, we will focus on the simulation parts, which are shown on the left side in the following diagram, a lot of effort has been made to keep the structure of the simulator similar to the real system. Ideally, all high-level components used in a simulated environment could be run on the real platform without any major changes:

All components, used on real MAVs, can be simulated by Gazebo plugins and a physics engine. We have developed the simulator as a modular way of assembling MAVs, where an MAV consists of a body frame, a fixed number of rotors, which can be placed at specified locations, and several sensors, which can be attached to the body as required. Each rotor has its own motor dynamics, the parameters of which can be identified on a real MAV, a Firefly from Ascending Technology using recorded flight data. Similarly, several sensors, such as an Inertial Measurement Unit (IMU), a stereo camera, GPS, and sensors developed by the user can be attached to the body frame. Moreover, we will also implement noise models for the applied sensors to simulate realistic conditions.

We will discuss an implementation of a geometric controller with a simple interface to facilitate the development of different control strategies. This provides access to various levels of commands, such as angular rates, attitude, or position control.

One of the important components is the state estimation, which is used to obtain information about the state of the MAV at a high rate. Although state estimation is crucial on real MAVs, in the simulation state parameters such as position, orientation, linear, and the angular velocity of the MAV are directly provided by a Gazebo plugin.