Semantic SLAM with Geometric Constraints for Object Tracking in Dynamic Environments

Visual simultaneous localization and mapping (SLAM) in dynamic environments has received significant attention in recent years, and accurate segmentation of real dynamic objects is the key to enhancing the accuracy of pose estimation in such environments. In this study, we propose a visual SLAM approach based on ORB-SLAM3, namely dynamic multiple object tracking SLAM (DMOT-SLAM), which can accurately estimate the camera’s pose in dynamic environments while tracking the trajectories of moving objects. We introduce a spatial point correlation constraint and combine it with instance segmentation and epipolar constraint to identify dynamic objects. We integrate the proposed motion check method into DeepSort, an object tracking algorithm, to facilitate inter-frame tracking of dynamic objects. This integration not only enhances the stability of dynamic features detection but also enables the estimation of global motion trajectories for dynamic objects and the construction of object-level semi-dense semantic maps. We evaluate our approach on the public TUM, Bonn, and KITTI dataset, and the results show that our approach has a significant improvement over ORB-SLAM3 in dynamic scenes and performs better compared to other state-of-the-art SLAM approaches. Moreover, experiments in real-world scenarios further substantiate the effectiveness of our approach.

Multi-camera visual SLAM for autonomous navigation of micro aerial vehicles

Current visual simultaneous localization and mapping (SLAM) systems have demonstrated commendable efficacy in static environments. However, the presence of dynamic objects in real-world settings frequently leads to system discrepancies, significantly impairing the accuracy and robustness of SLAM systems. Conventional visual SLAM approaches typically utilize epipolar constraints to mitigate the impact of outliers; nevertheless, they encounter limitations when confronted with a substantial number of dynamic or planar moving objects. To tackle these issues, this paper introduces a novel dynamic visual SLAM system, termed DEG-SLAM. Initially, the system employs the YOLOv5 object detection network to identify dynamic objects, subsequently relaying the semantic information to the tracking module. During the tracking phase, both semantic information and epipolar constraints are leveraged to filter out dynamic feature points. To address the challenges posed by the malfunctioning of epipolar constraints in degenerate scenes, DEG-SLAM incorporates a degenerate constraint mechanism aimed at further eliminating dynamic feature points. Furthermore, a reprojection constraint has been introduced to enhance the filtering of absent dynamic feature points that lie outside the detection boxes. Experimental findings reveal that DEG-SLAM significantly improves accuracy and robustness when compared to traditional ORB-SLAM3 in dynamic environments. The performance benefits of DEG-SLAM are particularly evident in degenerate scenarios, thereby affirming its practicality and reliability in complex settings.

Although tremendous progress has been made in Simultaneous Localization and Mapping (SLAM), the scene rigidity assumption limits wide usage of visual SLAMs in the real-world environment of computer vision, smart robotics and augmented reality. To make SLAM more robust in dynamic environments, outliers on the dynamic objects, including unknown objects, need to be removed from tracking process. To address this challenge, we present a novel real-time visual SLAM system, KMOP-vSLAM, which adds the capability of unsupervised learning segmentation and human detection to reduce the drift error of tracking in indoor dynamic environments. An efficient geometric outlier detection method is proposed, using dynamic information of the previous frames as well as a novel probability model to judge moving objects with the help of geometric constraints and human detection. Outlier features belonging to moving objects are largely detected and removed from tracking. The well-known dataset, TUM, is used to evaluate tracking errors in dynamic scenes where people are walking around. Our approach yields a significantly lower trajectory error compared to state-of-the-art visual SLAMs using an RGB-D camera.

Visual Simultaneous Localization and Mapping (VSLAM) is a critical technology for intelligent mobile robots, enabling simultaneous environmental mapping and self-localization. However, reflective ground in indoor environments presents significant challenges to VSLAM systems, as it introduces noisy virtual landmarks that degrade both map accuracy and localization precision. To address these challenges, we propose a novel RGB-D VSLAM system specifically designed for these environments. Our approach begins with a carefully designed point-line-plane-based VSLAM framework, which mitigates the limitations caused by sparse point features in indoor settings. Then we analyze how reflective ground degrades VSLAM performance and propose a reflective ground feature removal strategy that integrates semantic and geometric information. To enhance robustness in highly reflective environments, we incorporate a monocular depth estimation network as a complementary module. Extensive experiments on public datasets demonstrate that our framework achieves state-of-the-art performance. And evaluations on an author-collected dataset highlight the system’s superior mapping and localization capabilities in reflective indoor environments. Furthermore, ablation studies validate the effectiveness of each proposed component, and latency tests confirm the system’s suitability for real-time applications.

Visual simultaneous localization and mapping (VSLAM) is one of the foremost principal technologies for intelligent robots to implement environment perception. Many research works have focused on proposing comprehensive and integrated systems based on the static environment assumption. However, the elements whose motion status changes frequently, namely short-term dynamic elements, can significantly affect the system performance. Therefore, it is extremely momentous to cope with short-term dynamic elements to make the VSLAM system more adaptable to dynamic scenes. This paper proposes a coarse-to-fine elimination strategy for short-term dynamic elements based on motion status check (MSC) and feature points update (FPU). First, an object detection module is designed to obtain semantic information and screen out the potential short-term dynamic elements. And then an MSC module is proposed to judge the true status of these elements and thus ultimately determine whether to eliminate them. In addition, an FPU module is introduced to update the extracted feature points according to calculating the dynamic region factor to improve the robustness of VSLAM system. Quantitative and qualitative experiments on two challenging public datasets are performed. The results demonstrate that our method effectively eliminates the influence of short-term dynamic elements and outperforms other state-of-the-art methods.

Simultaneous localization and map building (SLAM) is crucial in autonomous robot navigation. However, existing SLAM systems generally assume a static environment, which makes it difficult to cope with the interference caused by moving objects in dynamic scenes, affecting the system’s localization accuracy and robustness. To address this challenge, this paper proposes YKD-SLAM, a visual SLAM system for indoor dynamic environments, which is based on the ORB-SLAM2 framework and incorporates YOLOv8 target detection, RCF-KMeans (Region-ConstrainedFastK-Means), and epipolar geometric constraints to realize the accurate rejection of dynamic feature points and improve the localization performance in dynamic environments. YKD-SLAM first uses YOLOv8 to detect dynamic objects in the scene, generates a detection frame, optimizes the depth map through open operations, and performs multi-region segmentation of the region within the detection frame by combining RCF-KMeans. Subsequently, through the dynamic feature point rejection strategy based on epipolar geometric constraints, different regions in the detection frame are discriminated into dynamic and static regions, and the feature points in the dynamic region are rejected to improve the localization accuracy and robustness of the system in dynamic environments. The experimental results show that YKD-SLAM performs well in several dynamic scenes in the TUMRGB-D dataset. Compared with ORB-SLAM2, its ATE is reduced by 98.37%; compared with DynaSLAM, the system operation efficiency is improved by 95.35%. In addition, practical experiments conducted in indoor dynamic scenes further validate its potential in real applications.

A comprehensive overview of core modules in visual SLAM framework

Localization and mapping in a dynamic scene is a crucial problem for the indoor visual simultaneous localization and mapping (SLAM) system. Most existed visual odometry (VO) or SLAM systems are based on the assumption that the environment is static. The performance of a SLAM system may degenerate when it is operated in a severely dynamic environment. The assumption limits the applications of RGB-D SLAM in the dynamic environment. In this paper, we propose a workflow to segment the objects accurately, which will be marked as the potentially dynamic-object area based on the semantic information. A novel approach for motion detection and removal from the moving camera is introduced. We integrate the semantics-based motion detection and the segmentation approach with an RGB-D SLAM system. To evaluate the effectiveness of the proposed approach, we conduct the experiments on the challenging dynamic sequences of TUM-RGBD datasets. The experimental results suggest that our approach improves the accuracy of localization and outperforms the state-of-the-art dynamic-removal-based SLAM system in both severely dynamic and slightly dynamic scenes.

During the continual transitions from lab research to real-world applications of vision-based algorithms, there are significant challenges, e.g. the robustness to adapt to complex environments and the high demands of multi-task, multi-modal learning models. More specifically on visual simultaneous localization and mapping (visual SLAM) for mobile robots, two notable limitations are: (i) the drift issue during pose estimation, especially in dynamic environments, makes the positioning system unstable; (ii) the additional loop detection takes many computational resources for image retrieval and global feature geometry checking. This thesis explores the approaches of visual localization on unmanned ground vehicles by solving those limitations. The conventional feature extraction and matching pipeline for vision-based tasks involve three steps: (i) local feature detection and description, (ii) feature matching, and (iii) outlier rejection. The first contribution of the thesis is adding a feature selection and anticipation stage to reduce tracking drift. We explore the noise model of image features to select a subset of all the observed image features with the best ”contribution” during data association and pose estimation across multiple frames. Conventional SLAM algorithms take a strong assumption of scene rigidity, which limits the application under challenging environments. The second part of the thesis addresses the tough issue in dynamic environments with moving objects. We presented GMC, namely the motion clustering approach, a lightweight dynamic object filtering method. It can distinguish moving objects from static landmarks. Based on the theory of motion coherence within a particular image area, GMC could segment dynamic objects in 3D space. We can provide an efficient and robust correspondence algorithm that can extract dynamic objects from a static background with the method. In this way, we propose a dynamic SLAM system that is real-time and free from expensive GPU processors. In contrast to GMC, the thesis’s third part turns to an object-aware learning-based model for more general dynamic scenarios. We use object detection and tracking as points, lines, planes, etc. We utilize semantic information and extract sparse image features simultaneously to keep track of dynamic objects. The static background and different dynamic objects are jointly optimized in a newly developed bundle adjustment sliding window. The estimated 3D bounding boxes can provide more robust camera tracking and better scene understanding, and better map merging. The fourth part of the thesis leverages the emerging feature learning framework. It proposes a unified self-supervised model called LGDNet to generate both global and local image feature descriptors end-to-end. Global feature descriptors embed the whole image into a compact representation, leading to easier scene comparison. On the other hand, local features focus more on the local region similarities of some salient parts for structure from motion. Our proposed method can directly extract features together with descriptors that encode both local maximum responses and global context information, avoiding duplicate calculations based on different feature extraction criteria.

In recent years, the signihcant progress has been made in visual simultaneous localization and mapping(VSLAM). Many present geometric VSLAM systems rely on static and bright environment assumptions, which is not friendly to the generalization of VSLAM in the real world including a large number of challenging scenes. To cope with these challenges, a real-time and robust visual-inertial SLAM system was proposed, which integrates a neural network for moving object detection(MOD) and greatly reduces the negative influence of dynamic objects. We has performed an ablation study to validate the effectiveness and necessity of our proposal. In addition, empirical evaluations on typical datasets, as well as in some usual dynamic environments, show that our novel framework can favorably solve the tracking loss, yield pure point cloud and improve the accuracy of VSLAM.

Simultaneous Localization and Mapping (SLAM) plays an important role in the computer vision and robotics field. The traditional SLAM framework adopts a strong static world assumption for analysis convenience. How to cope with dynamic environments is of vital importance and attracts more attentions. Existing SLAM systems toward dynamic scenes either solely utilize semantic information, solely utilize geometry information, or naively combine the results from them in a loosely coupled way. In this paper, we present SOF-SLAM: Semantic Optical Flow SLAM, a visual semantic SLAM system toward dynamic environments, which is built on RGB-D mode of ORB-SLAM2. A new dynamic features detection approach called semantic optical flow is proposed, which is a kind of tightly coupled way and can fully take advantage of feature's dynamic characteristic hidden in semantic and geometry information to remove dynamic features effectively and reasonably. The pixel-wise semantic segmentation results generated by SegNet serve as mask in the proposed semantic optical flow to get a reliable fundamental matrix, which is then used to filter out the truly dynamic features. Only the remaining static features are reserved in the tracking and optimization module to achieve accurate camera pose estimation in dynamic environments. Experiments on public TUM RGB-D dataset and in real-world environment are conducted. Compared with ORB-SLAM2, the proposed SOF-SLAM achieves averagely 96.73% improvements in high-dynamic scenarios. It also outperforms the other four state-of-the-art SLAM systems which cope with the dynamic environments.

To solve the accurate positioning problem of mobile robots, simultaneous localization and mapping (SLAM) or visual odometry (VO) based on visual information are widely used. However, most visual SLAM or VO cannot meet the accuracy requirements in dynamic indoor environments. This paper proposes a robust visual odometry based on deep learning to eliminate feature points matching error. However, when a camera and dynamic objects are in relative motion, the frames of camera will produce ghosting, especially in high-dynamic environments, which bring additional positioning error; in view of this problem, a novel method based on the average optical flow value of the dynamic region is proposed to identify feature points of the ghosting, and then the feature points of the ghosting and dynamic region are removed. After the remaining feature points are matched, we use a non-linear optimization method to calculate the pose. The proposed algorithm is tested on TUM RGB-D dataset, and the results show that our VO improves the positioning accuracy than other robust SLAM or VO and is strongly robust especially in high-dynamic environments.

This paper reviews the latest research progress of visual SLAM in dynamic scenes. The improved algorithms dependent on deep learning and multi-sensor fusion are discussed. First of all, the challenges of traditional visual SLAM in dynamic environments are stated, especially the interference of high dynamic objects on system feature extraction, pose estimation and mapping. Then, the application of several object detection technologies such as YOLOv5 and YOLOv7 in dynamic visual SLAM was analyzed in detail. These methods significantly improve the positioning precision and robustness of the system by identifying and eliminating dynamic feature points in the scene. In addition, this paper explores the visual SLAM system combining semantic segmentation and geometric constraint optimization, and finally introduces the multi-sensor SLAM method combining IMU data, point cloud data and visual information. Experimental results show that these improved algorithms have a significant improvement in positioning accuracy and system robustness compared with traditional visual SLAM algorithms under TUM and other dynamic data sets.

Loop closure detection is the core part of the visual simultaneous localization and mapping (VSLAM) for autonomous robots. In the dynamic environments, loop closure detection turns to be a very difficult problem compared to the static scenario. This paper proposes a novel approach based on convolutional autoencoder neural network (CAENN) architecture for extracting images features, and then uses Euclidean loss for minimizing the difference between the extracted feature and the gist feature which has the ability of scene recognition. In order to improve the accuracy and recall rate for loop closure detection in the dynamic scenario, the perspective transformation and dynamic object are applied in the process of the construction of the training set. And by calculating the Manhattan distance between the two image feature vectors, the loop closure is accepted when the distance is smaller than the threshold. The experimental results demonstrate that the proposed method obtains better accuracy and recall rate compared to the commonly used gist feature method and has a lower cost on time and space compared to the BOW method and AlexNet method in loop closure detection.