Point cloud segmentation is crucial for robotic visual perception andenvironmental understanding, enabling applications such as robotic navigationand 3D reconstruction. However, handling the sparse and unordered nature ofpoint cloud data presents challenges for efficient and accurate segmentation.Inspired by the Mamba model's success in natural language processing, wepropose the Serialized Point Cloud Mamba Segmentation Model (Serialized PointMamba), which leverages a state-space model to dynamically compress sequences,reduce memory usage, and enhance computational efficiency. Serialized PointMamba integrates local-global modeling capabilities with linear complexity,achieving state-of-the-art performance on both indoor and outdoor datasets.This approach includes novel techniques such as staged point cloud sequencelearning, grid pooling, and Conditional Positional Encoding, facilitatingeffective segmentation across diverse point cloud tasks. Our method achieved76.8 mIoU on Scannet and 70.3 mIoU on S3DIS. In Scannetv2 instancesegmentation, it recorded 40.0 mAP. It also had the lowest latency andreasonable memory use, making it the SOTA among point semantic segmentationmodels based on mamba.