To address the reliance on manual operation and the insufficient recognition and positioning accuracy during the segment grabbing stage of shield tunnel segment erectors, this paper proposes a segment bolt recognition and localization method that integrates an improved YOLOv8n model (YOLOv8n-PBSM) with RGB-D depth information. In the YOLOv8n network, a small-target P2 detection layer is added to enhance the extraction capability for bolt detail features. The original neck network is replaced with a bidirectional feature pyramid network (BiFPN) to strengthen multi-scale feature fusion. A parameter-free three-dimensional attention mechanism (SimAM) is introduced to improve the feature focusing ability of the model in complex tunnel environments characterized by oil contamination and uneven lighting. The MPDIoU loss function is adopted to optimize the bounding box regression process, thereby improving the localization accuracy for segment bolts. In the recognition stage, the YOLOv8n-PBSM model is utilized to identify segment bolts and achieve segment type classification via image markers. In the localization stage, depth information from an RGB-D camera is fused to establish a mapping process from pixel coordinates to the robot base coordinate system, enabling the three-dimensional spatial coordinate calculation of segment bolts. Experimental results show that the improved model achieves a mean average precision (mAP@0.5) of 96.36%, which is 3.71 percentage points higher than the original YOLOv8n. In simulated tunnel construction environments, the average time required for a single grasping operation is 5 s, with the detection and localization module taking only 65 ms. The axial positioning errors are all less than 2 mm, and all 15 grasping trials were successful. The experimental results validate that the proposed method meets the practical operational requirements for both recognition accuracy and positioning precision for segment bolts, providing a reliable visual perception solution for the automated assembly of shield tunnel segments.