Prior Guided Dropout for Robust Visual Localization in Dynamic Environments

Robust robot localization with visually adaptive consensus filters in dynamic corridor environments

Current simultaneous localization and mapping (SLAM) algorithms perform well in static environments but easily fail in dynamic environments. Recent works introduce deep learning-based semantic information to SLAM systems to reduce the influence of dynamic objects. However, it is still challenging to apply a robust localization in dynamic environments for resource-restricted robots. This paper proposes a real-time RGB-D inertial odometry system for resource-restricted robots in dynamic environments named Dynamic-VINS. Three main threads run in parallel: object detection, feature tracking, and state optimization. The proposed Dynamic-VINS combines object detection and depth information for dynamic feature recognition and achieves performance comparable to semantic segmentation. Dynamic-VINS adopts grid-based feature detection and proposes a fast and efficient method to extract high-quality FAST feature points. IMU is applied to predict motion for feature tracking and moving consistency check. The proposed method is evaluated on both public datasets and real-world applications and shows competitive localization accuracy and robustness in dynamic environments. Yet, to the best of our knowledge, it is the best-performance real-time RGB-D inertial odometry for resource-restricted platforms in dynamic environments for now. The proposed system is open source at: <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/HITSZ-NRSL/Dynamic-VINS.git</uri>

The state-of-art simultaneous localization and mapping (SLAM) methods have been demonstrated of satisfactory results in static environments. However, most of the existing methods cannot be directly applied in dynamic environments to account for moving objects. To solve the practical problem of SLAM in dynamic environments, an RGB-D SLAM method is developed based on moving objects removal and dense map reconstruction. Specifically, deep learning-based instance segmentation is performed to obtain the semantic information which is furthermore combined with multiview geometry theory to detect possible moving objects. Kalman filter and feature fusion algorithms are further used for moving objects tracking to improve the detection accuracy. Then the camera pose is estimated using the static feature points after motion removal to realize reliable localization in dynamic environments. Out of concern for the infeasibility of navigation and perception with sparse feature maps, each-single-view of dense map is constructed using the RGB image and the depth map collected by the RGB-D camera after motion removal. The dense environmental map is finally reconstructed through the registration of point cloud of selected key image frames. Experiments are conducted on public datasets and real scene data to evaluate the performance of the proposed method. Compared with the state-of-art method ORB-SLAM2, the absolute trajectory error and the relative pose error obtained by the proposed method are significantly reduced by at least 91%, and the localization accuracy reaches 0.001 m.

Visual localization has been well studied in recent decades and applied in many fields as a fundamental capability in robotics. However, the success of the state of the arts usually builds on the assumption that the environment is static. In dynamic scenarios where moving objects are present, the performance of the existing visual localization systems degrades a lot due to the disturbance of the dynamic factors. To address this problem, we propose a novel sparse motion removal (SMR) model that detects the dynamic and static regions for an input frame based on a Bayesian framework. The similarity between the consecutive frames and the difference between the current frame and the reference frame are both considered to reduce the detection uncertainty. After the detection process is finished, the dynamic regions are eliminated while the static ones are fed into a feature-based visual simultaneous localization and mapping (SLAM) system for further visual localization. To verify the proposed method, both qualitative and quantitative experiments are performed and the experimental results have demonstrated that the proposed model can significantly improve the accuracy and robustness for visual localization in dynamic environments.<;/p><;p><;i>Note to Practitioners<;/i>-This article was motivated by the visual localization problem in dynamic environments. Visual localization is well applied in many robotic fields such as path planning and exploration as the basic capability for a mobile robot. In the GPS-denied environments, one robot needs to localize itself through perceiving the unknown environment based on a visual sensor. In real-world scenes, the existence of the moving objects will significantly degrade the localization accuracy, which makes the robot implementation unreliable. In this article, an SMR model is designed to handle this problem. Once receiving a frame, the proposed model divides it into dynamic and static regions through a Bayesian framework. The dynamic regions are eliminated, while the static ones are maintained and fed into a feature-based visual SLAM system for further visual localization. The proposed method greatly improves the localization accuracy in dynamic environments and guarantees the robustness for robotic implementation.

Accurate localization and reliable mapping is essential for autonomous navigation of robots. As one of the core technologies for autonomous navigation, Simultaneous Localization and Mapping (SLAM) has attracted widespread attention in recent decades. Based on vision or LiDAR sensors, great efforts have been devoted to achieving real-time SLAM that can support a robot’s state estimation. However, most of the mature SLAM methods generally work under the assumption that the environment is static, while in dynamic environments they will yield degenerate performance or even fail. In this paper, first we quantitatively evaluate the performance of the state-of-the-art LiDAR-based SLAMs taking into account different pattens of moving objects in the environment. Through semi-physical simulation, we observed that the shape, size, and distribution of moving objects all can impact the performance of SLAM significantly, and obtained instructive investigation results by quantitative comparison between LOAM and LeGO-LOAM. Secondly, based on the above investigation, a novel approach named EMO to eliminating the moving objects for SLAM fusing LiDAR and mmW-radar is proposed, towards improving the accuracy and robustness of state estimation. The method fully uses the advantages of different characteristics of two sensors to realize the fusion of sensor information with two different resolutions. The moving objects can be efficiently detected based on Doppler effect by radar, accurately segmented and localized by LiDAR, then filtered out from the point clouds through data association and accurate synchronized in time and space. Finally, the point clouds representing the static environment are used as the input of SLAM. The proposed approach is evaluated through experiments using both semi-physical simulation and real-world datasets. The results demonstrate the effectiveness of the method at improving SLAM performance in accuracy (decrease by 30% at least in absolute position error) and robustness in dynamic environments.

Existing visual SLAM (vSLAM) systems fail to perform well in dynamic environments as they cannot effectively ignore moving objects during pose estimation and mapping. We propose a lightweight approach to improve the robustness of existing feature based RGB-D and stereo vSLAM by accurately removing dynamic outliers in the scene that contribute to failures in pose estimation and mapping. First, a novel motion state detection algorithm using the depth and feature flow information is presented to identify regions in the scene with high moving probability. This information is then fused with semantic cues via a probability framework to enable accurate and robust moving object extraction to retain the useful features for pose estimation and mapping. To reduce the computational complexity of extracting semantic information in every frame, we propose to extract semantics only on keyframes with significant changes in image content. Semantic propagation is used to compensate for the changes in the intermediate frames (i.e., non-keyframes). This is achieved by computing the dense transformation map using the available feature flow vectors. The proposed techniques can be integrated into existing vSLAM systems to increase their robustness in dynamic environments without incurring much computation cost. Our work highlights the importance of distinguishing between motion states of potential moving objects for vSLAM in highly dynamic environments. We provide extensive experimental results on four well-known RGB-D and stereo datasets to show that the proposed technique outperforms existing vSLAM methods in indoor and outdoor environments under various dynamic scenarios including crowded scenes. We also perform our experiments on a low-cost embedded platform, i.e., Jetson TX1, to demonstrate the computational efficiency of our method.

Task allocation for Unmanned Aerial Vehicle (UAV) swarms is a complex challenge. Effective distribution of tasks among multiple UAVs necessitates the optimization of conflicting objectives, such as task balance, flight distance, cost, and target benefits. This multi-objective optimization problem is inherently intricate due to strong interdependencies, nonlinear relationships, and dynamic environments. Traditional methods, including heuristic algorithms, supervised machine learning, and basic reinforcement learning mechanisms, often suffer from low efficiency and strict requirements for data sources or supervision mechanisms. They also struggle to adapt to problems with large discrete action spaces. To address this challenge, this paper proposes a novel approach that integrates swarm intelligence simulation with reinforcement learning. The core of this approach is a reinforcement learning algorithm based on the Deep Deterministic Policy Gradient (DDPG) framework. A key innovation is a customized DDPG variant specifically designed to handle the large, discrete action spaces inherent in UAV task allocation. Additionally, a behavior tree model is employed to simulate UAV swarm interactions, providing a realistic and verifiable environment for evaluating task allocation strategies. This simulation generates a rich dataset for training and testing reinforcement learning algorithms. The combined approach efficiently addresses the multi-objective optimization problem, enabling optimal task distribution for numerous UAVs. Compared to existing heuristic methods, the proposed approach demonstrates superior computational efficiency and solution quality, while maintaining reasonable convergence stability and environmental adaptability. Simulation-based validation ensures the approach’s robustness in dynamic environments, a critical factor for practical applications. Overall, the integration of simulation and reinforcement learning provides a powerful framework for tackling the complex task allocation problem in UAV swarms.

In robotic pick-and-place systems, challenges such as occlusions, complex geometries, and dynamic environments require advanced solutions for accurate grip detection. Conventional methods often fail to meet the necessary flexibility and precision, necessitating more robust approaches. While the abstract initially references robotic systems, the primary focus of this paper is on the application of Graph Neural Networks and attention mechanisms for motion analysis in physical education settings, aimed at enhancing student motion evaluation. The methodology described here is not intended for robotic applications but for analysing human movement. This work introduces a hybrid method combining Global Average Pooling (GAP) and Feature Hierarchical Kernel-based Grasp Pose Detection (FHKGPD), addressing the limitations of traditional systems. Unlike previous approaches, this method ensures efficient global feature extraction and improved robustness in dynamic environments. The goal is to enhance predictive performance and system stability for robotic applications, advancing object handling capabilities. The experimental results, explaining the effectiveness of the developed method, show the model with a 4.2 % RME value; interestingly, the obtained metrics are remarkable, e.g., an accuracy of 98.9%, a precision of 96.4%, a recall of 97.8%, and an F1-score of 97.1 %. These results account for its excellent performance versus competing approaches, which makes it a more trustworthy and flexible alternative to address industrial robot applications in tough, real-world scenarios.

The exponential growth of data traffic in 6G drives the dense deployment of small cells. While spectrum reuse enhances capacity, it inevitably exacerbates inter-user and inter-cell interference. Consequently, power allocation is vital for maximizing the instantaneous downlink sum rate subject to transmit power constraints. However, traditional optimization methods, such as fractional programming, rely on global Channel State Information (CSI) and incur high computational complexity, rendering them impractical for dynamic, large scale networks. Although Deep Reinforcement Learning (DRL) offers a promising alternative, existing methods are limited by full CSI dependence, misalignment with instantaneous objectives, and inefficient exploration, hindering robustness in dynamic environments. To address these challenges, this study proposes an enhanced DRL-based power allocation framework termed Information-Gain Deep Deterministic Policy Gradient (IG-DDPG). First, a compact state representation is designed by retaining only the top-I strongest normalized interference channel gains relative to the serving link, while discarding weak and negligible interference sources. This selection rationale effectively preserves dominant interference information to maintain decision accuracy while significantly reducing the state space dimensionality to remove the dependency on global CSI. Second, a max-reward learning mechanism is incorporated into the Deep Deterministic Policy Gradient (DDPG) architecture to directly optimize instantaneous sum rate. Third, building upon the Intrinsic Curiosity Module (ICM), we reinterpret its prediction-error signal as an information-gain surrogate and integrate it as an intrinsic objective. A weighting parameter is introduced to balance the trade-off between the extrinsic instantaneous rate and the intrinsic information gain, guiding the agent toward dynamically informative state transitions. Extensive experiments demonstrate that IG-DDPG consistently outperforms the baseline DDPG algorithm. It improves the average sum rate by no less than 4.3%, increases the top 20% average sum rate by at least 2.4%, and reduces variance by up to 22.15%. These results confirm that IG-DDPG provides a scalable and robust solution for real-time interference management in 6G ultra-dense cellular networks.

To address inefficiency, high safety risks, and excessive labor costs in pallet loading within the tobacco logistics industry, this paper proposes an intelligent loading system for automated guided vehicles (AGVs) based on simultaneous localization and mapping (SLAM) and multi-sensor fusion. The system integrates LiDAR, depth cameras, and inertial measurement units (IMUs). Through data fusion and adaptive localization algorithms, it achieves high-precision mapping, positioning, and dynamic obstacle avoidance. Experimental results show that under conditions of a 10 cm vehicle offset and a 3°angular deviation, the system successfully navigates with a 100% success rate. Loading accuracy is maintained within 0.5– 1.7 cm, and the average loading time is reduced to 1.5 minutes, improving efficiency by 50% compared to manual operation. Moreover, the system demonstrates strong robustness in dynamic environments, effectively handling the challenges of complex warehouse scenarios. The proposed approach enhances automation and safety in logistics operations and shows great potential for application in the tobacco industry and beyond.

Overlay topology plays an important role in P2P systems. Topology serves as a basis for achieving functions such as routing, searching and information dissemination, and it has a major impact on their efficiency, cost and robustness. Furthermore, the solution to problems such as sorting and clustering of nodes can also be interpreted as a topology. In this paper we propose a generic protocol, T-MAN, for constructing and maintaining a large class of topologies. In the proposed framework, a topology is defined with the help of a ranking function. The nodes participating in the protocol can use this ranking function to order any set of other nodes according to preference for choosing them as a neighbor. This simple abstraction makes it possible to control the self-organization process of topologies in a straightforward, intuitive and flexible manner. At the same time, the T-MAN protocol involves only local communication to increase the quality of the current set of neighbors of each node. We show that this bottom-up approach results in fast convergence and high robustness in dynamic environments. The protocol can be applied as a standalone solution as well as a component for recovery or bootstrapping of other protocols.KeywordsNetwork SizeBinary TreeRanking FunctionOverlay NetworkDistribute Hash TableThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Remote photoplethysmography (rPPG) is a non-contact technology that estimates physiological signals, such as Heart Rate (HR), by capturing subtle skin color changes caused by periodic blood volume variations using only a standard RGB camera. While cost-effective and convenient, it suffers from a fundamental limitation: performance degrades severely in dynamic environments due to susceptibility to noise, such as abrupt illumination changes or motion blur. This study presents a deep learning framework that combines two structural modifications to ensure robustness in dynamic environments, specifically modeling movement noise and illumination change noise. The proposed framework structurally cancels global disturbances, such as illumination changes or global motion, through a dual-branch pipeline that encodes the face and background in parallel after Video Color Magnification (VCM) and then performs differencing. Subsequently, it utilizes a structure that injects a Temporal Shift Module (TSM) into the Spatio-Temporal Feature Extraction (SSFE) block to preserve long- and short-term temporal correlations and smooth noise, even amidst short and irregular movements. We measured MAE, RMSE, and correlation on the standard dataset UBFC-rPPG under four noise conditions: clean, illumination change noise, Movement Noise, Both Noise and the real-world in-vehicle dataset MR-NIRP (Stationary and Driving). Experimental results showed that the proposed method achieved consistent error reduction and correlation improvement compared to the VS-Net baseline in the illumination change noise-only and combined noise environments (UBFC-rPPG) and in the high-noise driving scenario (MR-NIRP). It maintained competitive performance in motion-only noise. Conversely, a modest performance disadvantage was observed under clean conditions (UBFC) and quasi-clean stationary conditions (MR-NIRP), interpreted as a design trade-off focused on global noise cancellation and temporal smoothing. Ablation studies demonstrated that the dual-branch pipeline is the primary contributor under illumination change noise, while TSM is the key contributor under movement noise, and that the combination of both elements achieves optimal robustness in the most complex scenarios.

This paper presents a comprehensive AI-driven motion detection and analytics system using computer vision and machine learning techniques. The proposed approach combines traditional motion detection methods such as background subtraction with deep learning–based object detection and tracking to improve robustness in dynamic environments. The system is capable of detecting motion, classifying objects, tracking trajectories, and extracting meaningful analytics from video streams. Experimental evaluations demonstrate improved accuracy and reliability compared to conventional techniques, making the system suitable for surveillance, smart city, and intelligent monitoring applications.

Distributed Denial of Service (DDoS) attacks have emerged as one of the most pervasive and damaging threats to network security, disrupting services and incurring substantial financial costs. Machine learning (ML) has been widely explored as a potential solution to enhance DDoS detection and mitigation. This systematic review evaluates the effectiveness of ML techniques in detecting DDoS attacks, synthesizing findings from studies published between 2004 and 2024. The review analyzes models such as Random Forest, Support Vector Machines (SVM), and K-Nearest Neighbors (K-NN) based on key performance metrics like accuracy, precision, recall, and F1-score. A comprehensive search of multiple databases, including Web of Science, IEEE, Scopus, ScienceDirect, and Google Scholar, resulted in 19 studies that met inclusion criteria. The findings show that ensemble methods, particularly Random Forest, consistently outperformed other models in terms of detection rates, mainly due to their ability to handle large feature sets and reduce overfitting. Support Vector Machines also performed well in specific scenarios. However, their effectiveness was sometimes limited by computational complexity and the dataset size. K-Nearest Neighbors showed mixed results, depending on the nature of the attack patterns. This review emphasizes the potential of ensemble learning approaches for DDoS detection, demonstrating their robustness in dynamic environments. The review identifies key gaps in the existing research, including the need for better feature selection and exploring deep learning techniques to enhance DDoS detection accuracy and adaptability further. This study contributes valuable insights into the strengths and limitations of ML-based DDoS detection models, offering a foundation for future advancements in the field. It underscores the importance of continued research into hybrid and deep learning models to address the evolving and increasingly sophisticated nature of DDoS attacks in real-world applications.

A hierarchical framework for spectro-temporal feature extraction