CISC: coding curvature and intensity relations for loop closure detection

Loop closure detection is a critical and challenging task in autonomous driving, especially in enclosed indoor environ- ments like underground parking garages. In these scenarios, the limitations of traditional open-source algorithms lead to reduced accuracy and reliability, which presents significant challenges in practical applications. This study addresses these issues by proposing an innovative loop detection method that enhances performance through the fusion of point cloud curvature and intensity similarity. First, we introduce a novel global descriptor that integrates both geometric and intensity features, providing a more comprehensive representation of the surrounding environment. Second, recognizing the sequential correlation between elements of the global descriptor, we improve the construction of the ring key and apply it to consistency verification, thereby significantly enhancing the accuracy and efficiency of loop detection. Compared to current state-of-the-art algorithm frameworks, experimental results demonstrate that our proposed method achieves higher accuracy, better recall rates, and faster processing speeds in enclosed indoor scenarios. The successful application of this method offers a viable solution to the challenges of loop detection in closed environments within autonomous driving systems, presenting significant potential for practical implementation.