Developing intelligent robotic helpers for underground engineering presents significant hurdles, particularly with regard to autonomous navigation in areas devoid of GPS signals. Although drones using GPS navigation in open areas are sophisticated and widely used, when GPS signals are poor or in areas with subterranean space, drones become blind and lose their benefits. In this study, a thorough foundation for autonomous piloting algorithms was put forward in order to create an autonomous drone, which can pilot autonomously in surroundings without external referrals. The technological framework combines sensor fusion, machine learning, path planning, simultaneous localization and mapping (SLAM), and real-time communication. First, data from LiDAR, ultrasonic sensors, and inertial measurement units (IMUs) are first combined using sensor fusion techniques to generate a 3D virtual model of the surrounding area. Real-time mapping and locating are facilitated by SLAM algorithms. Second, machine learning models use sensor data to inform piloting decisions, while computer vision algorithms facilitate obstacle detection and recognition. Third, using the virtual model’s observed impediments and predetermined mission objectives, path planning algorithms create safe and effective flying routes. Hardware and software were created to accompany a drone for test and demonstration. Through a loop of testing, validation, redundancy, and fail-safes, the autonomous piloting algorithms were integrated and improved. The drone equipped with the autonomous piloting algorithms was able to navigate in subterranean space on its own with no GPS navigation after trial and error. The capability and dependability of autonomous drone systems for inspection support in such conditions are to be improved through ongoing research and development.