Existing methods for place recognition in 3D point clouds either ignore partial structure information by converting 3D scans to 2D images or construct constrained bag-of-words (BoW) representations reliant on specific feature extraction algorithms. In this paper, we propose a novel method based on incremental and general bag of words. Incorporating an adaptable keypoint and 3D local feature extraction method, we employ an incremental BoW model that is updated regularly. This enables a coarse-to-fine candidate selection from the database. And a revisit can be identified following geometric verification. In addition, we propose a new supplementary metric that addresses the leaving-out issue of the conventional metric, enhancing the identification of true loops. Employing a state-of-the-art (SOTA) keypoint and feature extraction algorithm, we evaluate our method as well as SOTA place recognition methods using diverse datasets with varying qualities. Experimental results demonstrate that our method outperforms the baselines across all three datasets, showcasing robust performance and notable generalization capabilities.

Ground surfaces are often carefully designed and engineered with various textures to fit the functionalities of human environments and thus could contain rich context information for smart wearables. Ground surface detection could power a wide array of applications including activity recognition, mobile health, and context-aware computing, and potentially provide an additional channel of information for many existing kinesiology approaches such as gait analysis. To facilitate the detection of ground surfaces, we present LaserShoes, a texture-sensing-enabled system using laser speckle imaging that can be retrofitted to shoes. Our system captures videos of speckle patterns induced on ground surfaces and uses pre-processing to identify ideal images with clear speckle patterns collected when users’ feet are in contact with ground surfaces. We demonstrated our technique with a ResNet-18 model and achieved real-time inference. We conducted an evaluation in different conditions and demonstrated results that verified the feasibility.