Research/ PLANETARY ROBOTICS
Integrating Deep Learning and Intelligent Path Planning for Autonomous Lunar Rover Navigation Using Chandrayaan-2 High-Resolution Imagery
Jay Prakash Kumar, Kumari Aradhya, Arpita Kumari, Shreyas N G, Farhana Kausar
Motion Planning & NavigationOpenAlex topic: Robotic Path Planning Algorithms · Learning-based MethodsOpenAlex topic: Advanced Neural Network Applications · Perception & State EstimationOpenAlex topic: Robotics and Sensor-Based Localization
Chandrayaannamed in title, abstract
The lunar south pole features craters that are perpetually shaded, harboring water ice, alongside regions with nearly continuous illumination. These characteristics present significant exploring opportunities. However, there are several navigational difficulties because to the complicated topography, which includes steep slopes, thick crater fields, and shifting sunlight. This study offers a straightforward method for locating secure paths by combining artificial intelligence with Chandrayaan-2 imagery. The technique achieves a precision of 0.7940, a recall of 0.8057, and a mAP@0.5 of 0.8928 by utilizing YOLOv8 deep learning for real-time crater identification. Algorithms for coordinate transformation convert identified risks into rovers-compatible coordinates.