Construction of Variable-Resolution Maps and Incremental Localization Techniques in Extreme Environments

In the field of autonomous driving, map construction is an extremely important technological method that provides a solid foundation for subsequent tasks such as navigation and planning. Therefore, map construction and localization have always been the key and difficult points in the field of autonomous driving. The objective of this study is to use the sensor characteristics of millimeter wave radar to achieve different levels of positioning accuracy based on the complexity of the environment in specific environments such as haze, rain, snow, and smoke, thus achieving high-reliability positioning and navigation. This study proposes a new method for map creation and navigation, which achieves variable-resolution positioning based on the complexity of the environment. This method reduces the computational complexity of map construction and maintenance and reduces the storage space of the map. The method proposed in this article significantly reduces the positioning error. For example, when k=2, the error is 26.13. However, after incorporating sequential constraints, the error decreases to 4.25. This clearly demonstrates the effectiveness of the method.