Automatic take-off landing quadcopter control on dynamics platform

SAR (Search and Rescue) become an interesting topic discussed worldwide because of its urgency to save humans life. Recent natural disaster in Palu, Central Sulawesi Indonesia demonstrates the lack of mobility which can cause a problem for the SAR team reaching the victim within the isolated area. Uneasy job for the SAR team become more challenging in time response and in finding and saving disaster victims. Consider the fact, there is a need to develop more efficient and effective way to search and rescue the survivors. The using of Quadcopter can answer the problems. Unfortunately the Quadcopter still conventionally controlled by a remote and further it has to be on static ground whenever its take off or landing. This paper reports a Quadcopter fitted with dynamical autonomous take off and landing systems. Thus it can carry out the advantage from a tactical SAR vehicles or other moving platforms. Using object detection approach, The navigation system in an image-based control quadcopter or called visual servoing is the use of machine vision as a control of the closed loop position for movement. Images can be detected using YOLOv3 Real Time Object Detector, and tracks it by estimating the motion of objects in successive video frames. By using 3D reconstruction method the distance and position of the designated target and the quadcopter could be determined. Then the estimation of the relative position of the object is used as input for the control system on the quadcopter using PID controller. The controller commands the quadcopter to approach the target, while the image processing checks the relative position between the two. If the position satisfies the minimum landing parameter, then the controller commands a landing. While for take off we implement the controller to maintain its equilibrium and sufficient power to throttle.