Designing an Intelligent Pathfinder Drone withDeep Learning and Fuzzy Logic

Drones are flying robots with many structures that can fly unmanned and can composite with some degree of intelligence to help them make appropriate decisions when facing new situations. Also, with the progress of artificial intelligence (AI) in recent years, its application can be seen in most other branches. This paper presents the design of such an intelligent drone that is capable of doing proper maneuvering when it is flying over roads. specifically, we used a deep learning network to train it by a big dataset of images from nature and urban roads gathered by ۳ cameras and we got help from fuzzy logic to gain the best angle and speed that keep the drone in the path. The goal is to improve the accuracy of learning by extending the dataset and the network compared to previous works and decreasing the complexity by fuzzy logic. We used this method on the hexacopter on the road outside the city at low altitude which had good results.

کلیدواژه ها:

UAV ، Deep Learning ، Fuzzy Logic ، Neural Networks ، Auto Flight ، path finder ، autonomous