No abstract

No abstract

This article is a continuation of our previously published work that presented a comparison of nine attitude quaternion-based controllers of the quadrotor in simulation environment. In this article, the best three controllers were implemented into the real quadrotor. Namely proportional derivative (PD), linear quadratic regulator (LQR) and backstepping quaternion-based control techniques were evaluated. As a suitable test stand was not available on the basis of literature analysis, the article also outlines the requirements and the development of a new innovative test stand. In order to provide a comprehensive overview, the hardware and software that was used is also presented in the article. The main contribution of this article is a performance comparison of the controllers, which was based on absolute quaternion (positioning) error and energy consumption.The quadrotor is coupled and underactuated system (i.e., six degrees of freedom are controlled only by four actuators). The translational motion of the quadrotor depends on the attitude of the quadrotor as can be seen from the equations in [9]. By this means, the quadrotor is a cascade system [10]. Therefore, a cascade structure of a controller is usually applied to control position and attitude of the quadrotor. Various controllers based on classic or modern control theory were already designed and verified. The non-linear or linear model of a quadrotor is used depending on the chosen control method [11].The linear model of a quadrotor is achieved by linearization of the non-linear model around a hovering point. Controllers using the linearized model generally perform well around this hovering point. When the quadrotor goes away from the linearized point, the performance may worsen. Furthermore, the input saturation can cause control failure when large rapid maneuvers are required [3,5,12,13]. The main advantages of linear controllers are simplicity and ease of implementation to a real platform. The disadvantage of this approach is the use of the linearized model of the quadrotor during the process of designing a controller. Commonly used linear controllers are a linear quadratic regulator (LQR) and a proportional derivative (PD) controller [14][15][16].Another group of controllers consists of a wide variety of non-linear controllers. The serious disadvantage of these controllers is the complexity that prevents the wide adoption of non-linear controllers in real applications. Among non-linear methods, the backstepping control technique based on the Lyapunov function is widely adopted due to its systematic design and an intuitive approach [17]. The proposed control law is based on the compensation of non-linear forces or torques depending on whether an attitude or position controller is being designed. Applying Lyapunov stability analysis proves that the closed-loop system is asymptotically stable. This approach was used to stabilize the quadrotor in [3,5,12,[18][19][20][21].A backstepping-based inverse optimal attitude controller (BIOAC) was derived in [3] ta...

This article deals with the localization of an iRobot Create robotic platform using an inertial measurement unit (IMU). The used IMU consists of a gyroscope and accelerometer. The implementation of the IMU in the robotic platform allows the use of complex localization methods. We compare results of experimental measurement using odometry, gyroscope, accelerometer and their combination. In order to fuse sensor signals, a Complementary and a Kalman filter are designed and tuned. Various methods are used to verify results, such as UMBmark and rotation test. The combination of the IMU and mentioned filters provides better results of the localization in comparison with the odometry.

No abstract

No abstract