Advancement in Autonomous Navigation in Space through Artificial Intelligence: A Systematic Review

Abstract

Since the first trip to Mars, rovers have been used to conduct scientific experiments. The use of rovers continues to date due to Mars being uninhabitable for humans at this state and time. Despite the invention of Mars rovers, many limitations have been challenging space exploration, even with improvements in the Curiosity and Perseverance rovers. This study aimed to investigate the advancement in autonomous navigation in space through artificial intelligence. The study used qualitative, desktop, and systematic review research designs. The study's objectives were to determine the types of autonomous navigation technologies in space and examine the role of artificial intelligence in autonomous navigation technologies. Data was collected from secondary sources, which were purely open-access online journals. A total of 61 journals were searched as the target population. The study then conducted judgmental sampling using 9 out of 61 journals. The Preferred Reporting Items for Systematic Reviews and meta-analysis guidelines guided the sampling process. The study used an extraction form to collect data. The study found that star tracker navigation reduces positioning errors and environmental disturbances from sand and dust storms. It was also established that LOS navigation and star trackers ensured highly accurate navigation in challenging conditions for path-planning accuracy. Selecting the Next Coordinate, Obtaining Coordinates of Target Points, Generating Additional Coordinates, and optimising planned paths reduce travel time and energy consumption. On-orbit servicing robotics can extend rover missions up to 10 years. It was also found that 3D OctoMaps improves rovers' navigation accuracy. In addition, the study revealed that unsupervised homography networks, state recursive models, and inertial measurement models reduce localisation errors. Simultaneous Localization and Mapping can improve navigation accuracy and reduce mission delays. The use of RL navigation systems reduces mission delays. It was also found that Convolutional Neural network algorithms advance autonomous navigation accuracy in terms of terrain classification. The study also established that the Nvidia Jetson Nano AI controller and Rapidly-exploring Random Tree (RRT) in MATLAB reduce travel time and energy consumption. Finally, it was found that artificial neural networks advance the accuracy of terrain classification and lower navigation errors.