Algorithm of Air-Ground Localizing and Orienting Based on Semantic Information of Road Marking

The use of air-ground coordination system in urban environment can significantly increase the perception ability of the system, meanwhile, conversion of air-ground perspective is the key problem that restricts the realization of autonomous cooperation between UAV (Unmanned Aerial Vehicle) and UGV (unmanned ground vehicle). Considering the characteristics of road traffic signs in the urban environment, this paper proposes an air-ground view conversion method based on the semantic information of road traffic signs. Firstly, we preprocess the aerial view images and ground view images, establish semantic segment layout descriptors from the semantically segmented images according to the encoding rules. Secondly, the descriptors can be matched with each other to solve the homography matrix and calculate the relative position and attitude between the UAV and UGV by decomposing the matrix. Finally, we conduct experiments in the Gazebo simulation environment. The experimental results verify the accuracy and practicability of the method proposed in this paper and the problems and limitations are analyzed.