Odometry with Real-Time Depth Completion for UAV Location in Denied Environment

With the advent of the information age, the application of unmanned aerial vehicles (UAVs) has become increasingly frequent, and the combat scenarios have become more diverse. How to enable UAVs to achieve precise localization in a denial environment has become a focal point of interest. Consequently, Simultaneous Localization and Mapping (SLAM) technology is widely used on UAVs. Within SLAM technology, depth information provides a critical guarantee for precise localization. Depth completion, as an important issue in computer vision, leverages sparse depth data obtained from sensors and combines it with semantic information provided by color images to acquire dense depth information of the scene. Following this idea, this paper improves a multimodal fusion SLAM algorithm and integrating an improved depth completion algorithm to enhance the localization accuracy of the SLAM system and expand its mapping ability.