Precise localization for achieving next-generation autonomous navigation: State-of-the-art, taxonomy and future prospects

Abstract Achieving full autonomy in navigation is a complicated problem. The most widely used solution takes up the modular framework for sensing and information processing such as perception, mapping, control, planning and decision making. However, this approach misses the capability of environmental understanding. Hence, to achieve full autonomy in navigation a computing model with self-learning capability inspired by biological intelligence such as memorizing, inferring and experience update is essential for dynamic and noisy environments. Recent advanced sensing, communication and hardware miniaturization technologies achieved few autonomous operations in commercial systems but the full autonomy has not been attained yet. In this paper, the effect of precise and accurate localization for autonomous navigation technologies is extensively studied and the problems and limitations of the related algorithms are analyzed. The major limitations for precise localization are computational complexity, sensor noise and communication delays. These limitations further reduce perception and planning capabilities of autonomous navigation systems. From this study, the future prospects are outlined to achieve a higher level of autonomy by precise localization.