SLIDING MODE CONTROL OF A QUAD-COPTER FOR AUTONOMOUS TRAJECTORY TRACKING

Unmanned air vehicles or drones have become ubiquitous in our daily lives; they are deployed in performing many tasks from dangerous military missions to simple recreation activities. One air vehicle that has become very popular is the quad-copter driven by four vertical and parallel propellers. Today quad-copters are deployed in many video recording and remote monitoring almost everywhere in the world. One area of interest for quad-copters has been in farming operations; these vehicles are used in farming operations for not only aerial monitoring of soil nitrogen levels but many other farm monitoring operations. One common aspect of most quad-copters is that they are teleoperated by the user, i.e., most of them are not yet fully autonomous. There must be a remote pilot who is connected to the quad-copter by a video link so that he/she can control the maneuver of the vehicle along the intended path. This paper intends to show that a quad-copter can be programmed to run autonomously along a predetermined trajectory by using sliding mode control strategy. Since trajectories in most farms are clearly well known in advance, then they can be programmed into the controller for the quadcopter to autonomously track. The design process involves using the intended trajectory to define the 3-D sliding surface and then letting the quad-copter controller switch about that surface while keeping the vehicle in the target trajectory. The workspace is defined as a 3-D space where the sliding surface is defined by fitting weighted spline functions on the coordinates of the intended trajectory to define the stable sliding surface whose stability lever increases as the vehicle moves towards the target point. Preliminary results compare the trajectories followed by the quad-copter and the intended trajectories by using the mean square deviation. As would be expected, the performance depends heavily on the speed of the quad-copter; higher speeds on sharp curvature are associated with large tracking errors than low speeds on similar curvatures, while the performance on straight line paths was considerably good. This is most likely due to the switching speed because it seems that higher speeds should be associated with higher switching speeds also. The future work intends to study if parameterizing the 3-D splines using speed and time can improve the tracking performance where the switching rate will be made to be proportional to the number of spline functions that define the trajectory irrespective of the speed of the quad-copter.