This work presents a technique of collision avoidance for Unmanned Aerial Vehicle (UAV) using occupied grids algorithm. The proposed algorithm is utilizing bounding box based convolutional neural network (CNN). Image frame from UAV camera is divided i...
This work presents a technique of collision avoidance for Unmanned Aerial Vehicle (UAV) using occupied grids algorithm. The proposed algorithm is utilizing bounding box based convolutional neural network (CNN). Image frame from UAV camera is divided into grids. Occupied grids by bounding box defines the distance and position of static also dynamic object around UAV in real-time. A flight trajectory, velocity and direction of UAV are presented as performance matrix. Experimental results show that proposed algorithm successfully directs UAV to avoid collision with object by decreasing its velocity and changing direction.