Simultaneous localization and mapping (SLAM) is a fundamental function of intelligent robots. To reduce the influence of dynamic objects on SLAM in dynamic environments, this study pro-poses a visual SLAM based on sequential image segmentation, referred to as SIIS-SLAM. Based on ORB-SLAM3, SIIS-SLAM integrates the sequential image instance segmentation and optical flow dynamic detection module. The sequential image segmentation module is designed to eliminate the effectiveness of dynamic objects in the estimation of relative pose between sequential frames. Specifically, based on the coarse relative pose estimated by ORB-SLAM3 and the box coordinates of instances detected by Mask R-CNN, the sequential image segmentation module effectively improves the speed and accuracy of instance segmentation. Dynamic objects can be effectively detected by combining the instance segmentation results and optical flow module. Filtering the feature points in dynamic objects can improve the accuracy and robustness of SLAM. Experimental results demonstrate that SIIS-SLAM achieves the better accuracy in dynamic environments compared to ORB SLAM3 and other advanced methods.

Visual localization is a core part of many computer vision and geospatial perception applications; however, the ever-changing time phase and environment present challenges. Moreover, the ever-enriching spatial data types and sensors create new conditions for visual localization. Based on the prior 3D model and the location sensor, the current study proposes a visual localization method using semantic information. This method integrates panoptic segmentation and the matching network to refine the sensor’s position and orientation and complete the target perception. First, the panoptic segmentation and the match network are used together to segment and match the 3D- model-rendered image and the truth image. The matching results are then optimized based on the semantic results. Second, the semantic consistency score is introduced in the RANSAC process to estimate the optimal 6 degree-of-freedom (6DOF) pose. In the final stage, the estimated 6DOF pose, the instance segmentation results, and the depth information are used to locate the target. Experimental results show that the proposed method is a significant improvement on advanced methods for the long-term visual localization benchmark dataset. Additionally, the proposed method is seen to provide improved localization accuracy and is capable of accurately perceiving the target for self-collected data.

No abstract