PMF-SLAM: Pose-guided and Multi-scale Feature Interaction Based Semantic SLAM for Autonomous Wheel Loader

Donglin Zhu,Zhongli Wang,Tengfei Lu,Xin Jiang

DOI: https://doi.org/10.1109/jsen.2024.3352012

IF: 4.3

2024-01-01

IEEE Sensors Journal

Abstract:In a dynamic environment, semantic information can assist the simultaneous localization and mapping (SLAM) system in eliminating dynamic point interference. However, most 3-D semantic segmentation methods are computationally expensive and also have low segmentation accuracy for both distant and small objects. We propose a pose-guided and multiscale feature (PMF)-SLAM method to fully exploit the interaction between 3-D semantic segmentation and SLAM and achieve efficient scene perception. The PMF-SLAM system includes three parts: a multiscale feature based segmantation network (MSF-SegNet) model, an interactive SLAM module, and a pose-guiding segmentation module. To improve the accuracy of distant and small objects, MSF-SegNet merges point-wise global features and voxel-wise local features from two branches by a designed symmetrical sparse convolution structure. In the interactive SLAM module, the coarse-to-fine registration method based on semantic information completes the estimation of the pose. To implement the interaction between segmentation and SLAM, the pose-guiding segmentation module was built to assist the segmentation thread in improving inference efficiency and ensuring segmentation consistency over time. Extensive experiments including both local experiments and the nuScenes dataset test have been conducted to validate the performance of the proposed method. Our method achieves better accuracy than multiple segmentation algorithms, significantly improving the segmentation performance of distant and small objects. The trajectory estimation accuracy is better than multiple SLAM algorithms. The code is available at https://github.com/haroldgt/MSF-SegNet.

engineering, electrical & electronic,instruments & instrumentation,physics, applied