Vision-Based Tracking Control of Quadrotor With Backstepping Sliding Mode Control

Abstract

Vision-based quadrotor will be a good carrier for big data. This paper investigates the quadrotor tracking control by designing an adaptive sliding mode controller based on the backstepping technique with the advantages of simplicity in design and ease of application. A sliding mode controller is first developed to ensure fast convergence speed with the desired reference, and then the backstepping technique is used until the desired reference trajectory is achieved and finally the appropriate control laws are obtained. In order to achieve the precise and fast localization of a quadrotor, a popular visual odometry algorithm is applied to gathering good position information required in motion estimation. We employ Kalman filter for sensor data fusion and state estimation. Gazebo is applied by creating a 3D dynamic environment to recreate the complex environment potentially encountered in the real world.

Keywords

• Backstepping
• quadrotor
• Kalman filter
• semi-direct monocular visual odometry
• sliding mode control
• tracking control