Referring Video Object Segmentation (RVOS) aims to segment the target objects specified in human instructions. Previous approaches typically rely on explicit human instructions that contain target categories or salient appearance descriptions. These approaches tend to fail when the instructions require temporal video understanding and complex relational reasoning. In this work, we present RViSeg, a reasoning-centric video object segmentation model that leverages the reasoning capability of Multi-modal Large Language Models (MLLM) to handle complex queries. The primary challenge lies in enabling MLLM to perform efficient pixel-level video perception. To tackle this challenge, we introduce a novel Spatial Token Merge (STM) module that consolidates lengthy video tokens into compact region-level clusters, while preserving essential spatial details. This structured representation enables MLLM to infer user intention by interleaving spatial and temporal visual information. Furthermore, we propose a Query-based Target Retrieval (QTR) module that utilizes learnable tokens as the target identity for mask prediction. By propagating these instance-specific tokens both intra-clip and inter-clip, our RViSeg effectively encodes object motion, ensuring spatio-temporal consistency in segmentation results. To facilitate training and evaluation, we construct InstructVideo, a single- and multiple-object reasoning video segmentation benchmark. Comprehensive experiments demonstrate the effectiveness of the proposed components.

In robotic navigation, maintaining precise pose estimation and navigation in complex and dynamic environments is crucial. However, environmental challenges such as smoke, tunnels, and adverse weather can significantly degrade the performance of single-sensor systems like LiDAR or GPS, compromising the overall stability and safety of autonomous robots. To address these challenges, we propose AF-RLIO: an adaptive fusion approach that integrates 4D millimeterwave radar, LiDAR, inertial measurement unit (IMU), and GPS to leverage the complementary strengths of these sensors for robust odometry estimation in complex environments. Our method consists of three key modules. Firstly, the pre-processing module utilizes radar data to assist LiDAR in removing dynamic points and determining when environmental conditions are degraded for LiDAR. Secondly, the dynamic-aware multimodal odometry selects appropriate point cloud data for scan-tomap matching and tightly couples it with the IMU using the Iterative Error State Kalman Filter. Lastly, the factor graph optimization module balances weights between odometry and GPS data, constructing a pose graph for optimization. The proposed approach has been evaluated on datasets and tested in real-world robotic environments, demonstrating its effectiveness and advantages over existing methods in challenging conditions such as smoke and tunnels. Furthermore, we open source our code at https://github.com/NeSC-IV/AF-RLIO.git to benefit the research community.

Multi-robot exploration in unknown environments is a fundamental task for multi-robot systems, which requires the coordination of the robots to avoid collisions and conflicts while performing task allocation. Existing exploration strategies improve the efficiency of multi-robot exploration by modeling the multi-robot task allocation problem as a variant of the multiple traveling salesman problem. However, this is computationally intensive and difficult to deploy on physical platforms. Hence, this paper develops a hybrid strategy for range-sensing multi-robot exploration with effective team coordination, enabling a larger team dispersion degree and higher exploration efficiency. In addition, we present a novel multi-robot exploration point detection method suitable for narrow and dynamic environments, effectively reducing exploration failure and incompleteness. The Gazebo simulations demonstrate better exploration efficiency and the least time cost of our exploration framework compared with state-of-the-art methods, and real-world experiments also validate the effectiveness. The code is released at https://github.com/NeSC-IV/DHC_ME.

In recent years, multi-modal semantic segmentation in autonomous driving has garnered significant attention due to its effectiveness under challenging lighting conditions. However, current segmentation approaches primarily focus on segmentation techniques without addressing the critical communication challenges inherent in internet of vehicles (IoV). Unlike traditional communications that transmit source data, semantic communications transmit only task-relevant semantic information, significantly reducing data traffic while ensuring the accuracy of task execution. This paper introduces a novel cooperative multi-modal semantic communication framework designed to enhance semantic segmentation in autonomous driving systems. By compressing redundant information and transmitting only essential semantic features, the proposed scheme enables continuous data transmission with drastically reduced data volume. Moreover, this scheme not only improves communication efficiency but also ensures reliable segmentation performance across diverse data modalities. Experimental results validate the effectiveness of proposed scheme, demonstrating its ability to achieve high compression ratios, robust segmentation performance, and re-silience to channel noise under varying lighting conditions.

Distributed localization systems enable nodes to determine their locations by exchanging information with neighboring nodes, without relying on centralized infrastructure. While it enhances scalability and robustness, random data loss during in formation transmission can significantly degrade the localization performance. In this paper, to mitigate the impact of random data loss, we propose a Loss-Robust distributed localization method based on the Distributed Iterative Localization algorithm (LR DILOC). In LR-DILOC, each node updates its location estimate by leveraging the most recent available location information from its neighboring nodes, thereby improving the utilization of available data. We theoretically analyze the convergence of LR-DILOC, demonstrating that LR-DILOC maintains accurate localization even in the presence of random data loss. Numerical results further validate the theoretical analysis, demonstrating that LR-DILOC achieves higher localization accuracy and exhibits stronger robustness under random data loss.

In object goal navigation tasks, the robot’s understanding of semantic relationships in the environment is a key factor in its ability to localize target objects. Previously, learning-based methods trained robots using 3D scene datasets to learn semantic relationships. However, these approaches perform poorly in new environments with unfamiliar semantic contexts. In this paper, we propose ChatNav which leverages the powerful knowledge summarizing and reasoning capabilities of a Large Language Model (LLM) for zero-shot inference of explicit semantic relationships. These relationships are further integrated into the navigation system for efficient localization of target objects. ChatNav employs a spatial object clustering algorithm to collect semantic clues and designs common-sense-based prompts for interacting with LLM. It then uses a gravity-repulsion model to convert inference results into heuristic factors for robust navigation decision-making. Our approach requires no additional training and can consistently obtain accurate semantic relationships from LLM, making it well-suited for navigating unknown environments. Experimental results demonstrate the outstanding navigation performance of our proposed method on the Gibson and HM3D datasets, surpassing the current state-of-the-art object goal navigation methods.

The integration of multi-sensor data for accurate and robust state estimation is a promising research area with various practical applications. In this paper, we propose an optimization-based fusion framework that combines the camera, 6-DoF IMU, and UWB sensors for accurate and robust real-time localization. Different from traditional localization strategies, the proposed framework includes a data preprocessing module to deal with the issue of time asynchrony among multi-sensor data and a back-end optimization process that relies on UWB loop closure detection to correct the localization error caused by the UWB measurement noise and VIO drift. Moreover, a state prediction model that takes into account all collected UWB data between two consecutive keyframes is proposed to further improve the localization performance. Experiments on public datasets and real-life scenarios demonstrate the efficiency and robustness of the proposed method.

In cooperative localization, the sharing of location information has raised the risk of the privacy breach. Geo-indistinguishability, as a formal notion of location privacy, can protect the user’s location privacy by adding random noise into his true location, while it degrades the localization accuracy inevitably. This article considers the tradeoff between the cooperative nodes’ privacy preservation level and the target’s localization accuracy. Since it incurs privacy cost for the cooperative nodes to report their location information, an incentive mechanism for the cooperative nodes to contribute their data is necessary. In this article, we propose a feasible incentive mechanism, named CLAP, based on contract theory to reward the cooperative nodes and ensure the expected localization accuracy. Specifically, we establish an optimization problem of minimizing the payment from the target, while guaranteeing the expected localization accuracy. By simplifying the constraints and transforming the original problem into a convex optimization problem, a closed-form solution is given. Extensive simulations evaluate the performance and validate the feasibility of our proposed contract-based incentive mechanism.

This article studies the intrusion detection problem for industrial control systems (ICSs) with repetitive machining under false data injection (FDI) attacks. A data-driven intrusion detection method is proposed based on both time- and frequency-domain analysis. The proposed method only utilizes the sensor measurements required in closed-loop control, and does not consume additional system resources or rely on the system model. In addition, features in time and frequency domain are extracted at the same time, having higher reliability than the intrusion detection methods which only utilize the features in time domain. After feature extraction, hidden Markov models (HMMs) are established by using the feature vectors under normal operating conditions of the ICS, and then the trained HMMs are utilized in real-time intrusion detection. Finally, experiments are carried out on a networked multiaxis engraving machine with FDI attacks. The experimental results show the effectiveness and superiority of the proposed intrusion detection method.

Duty-cycling techniques have been widely adopted to save energy for energy-constrained wireless sensor networks, while they also cause the sleep latency issue, especially in a multihop linear sensor network (LSN). So the duty-cycling and pipelined-forwarding (DCPF) techniques have been proposed to alleviate this issue. However, most of existing DCPF protocols have no effective scheme to handle the contention and interference among those proximately-located nodes which maintain the same sleep-wakeup schedule. As a result, the network performance degrades with low energy efficiency and high packet delivery latency, particularly when experiencing a heavy traffic load. To this end, this paper proposes an enhanced pipelined data collection (EPDC) MAC protocol for LSN. In EPDC, three algorithms are proposed to guarantee that those nodes located within the interference range of each other have mutually staggered sleep-wakeup schedules, so that the contention and interference among them can be eliminated. The extensive OP-NET simulations show that EPDC significantly outperforms an existing DCPF protocol in terms of packet delivery ratio, network throughput, packet delivery latency, and energy efficiency.

Existing distributed localization methods rarely consider the location privacy preservation problem, which however is nonnegligible. Regarding location privacy, typical solutions rely on a curious-but-honest model, requesting that all participants follow the rule. Different from the existing studies, both honest and dishonest models are considered in this article. We first propose a privacy-preserving distributed localization algorithm (PP-DILOC) by adopting a noise-adding mechanism under the curious-but-honest model. The performance of localization and privacy preservation of PP-DILOC are both theoretically analyzed. Then, in the presence of dishonest nodes, we propose a resilient PP-DILOC (RPP-DILOC), where a time-varying relax factor and an adversary detection procedure are added into PP-DILOC. Theoretical results provide sufficient conditions for the convergence of RPP-DILOC. The privacy levels and the localization performance in the absence/presence of dishonest nodes are evaluated through numerical and experimental results.

In wireless localization systems, location information with respect to anchors will be exposed to untrusted target or third party inevitably leading to location privacy leakage. Differential privacy based techniques can provide theoretical guarantee to privacy preservation, while such privacy preservation will degrade localization accuracy. We note that the quantitative relationship between localization accuracy and privacy level is still unclear. In this paper, we derive the Cramér-Rao lower bound (CRLB) about the target location when the anchors take a privacy preservation mechanism satisfying geo-indistinguishability, which is an application of differential privacy on Euclidean metric. The closed-form relationship between the target localization accuracy and the anchors' location privacy level is provided respectively for range-only and bearing-only localization. Numerical results in further verify our theoretical results.

To invade a cyber-physical system (CPS) successfully, hackers are prone to simultaneously launching multiple cyber attacks on different sensors in a CPS. However, little attention has been paid to the problem of detecting multiple cyber attacks up to now. Therefore, in this paper, we deal with the problem on how to efficiently detect multiple cyber attacks aiming at different sensors in CPSs. To achieve the goal of simultaneously detecting both the number of attacks and the attacked sensors, we formulate this problem via a random finite set (RFS) theory, and then apply an iterative RFS-based Bayesian filter and its approximation to solve the problem. Four numerical experiments with different attacks are provided, and the results have demonstrated the effectiveness of the RFS-based approach for the problem of multiple attacks detection in CPSs.

The increasing popularity of drones is bringing many public security and privacy breach issues, such as smuggling, intrusion, and illegal surveillance. Traditional approaches to detecting and localizing drones such as radar and computer vision incur high costs and hence are not desirable for large-scale applications. In this paper, we propose a cost-effective crowdsensing system named CEDAR to achieve such a goal. Specifically, we introduce a novel way of detecting drones by smartphones, exploiting the fact that most drones adopt Wi-Fi for communications with ground control stations. We design an efficient detection algorithm that takes advantage of historical Wi-Fi beacon information and MAC address encoding mechanisms used by drone manufacturers. Using received signal strength, we can also localize the detected drones. Further, to encourage participants' involvement, we design an incentive mechanism based on online auction that guarantees truthfulness and consumer sovereignty. CEDAR can be directly applied to multiple drone scenarios. We implement the system based on Android for the client and Spring, Spring MVC, and Mybatis (SSM) for the centralized platform that supports scalability and hierarchical structure, and enables the coordination between clients and the platform. We perform extensive experiments to validate our analysis. Particularly, the detection rate in the experiments reaches 86.7 percent even without any prior information about drones.

Unlike traditional multilateration techniques which usually require multiple anchor nodes to participate in localization process, this paper proposes a novel Single Anchor Localization (SAL) scheme employing target movement. The scheme requires the anchor node to only measure its distance to the target node and the target node to measure its moving distance along straight lines. Two algorithms, i.e., SAL based on two rectilinear movings and a triangular moving, are proposed depending on whether the target node also needs to measure its turning angle or not. The moving distance of the target can be estimated using human step or height, geographical indication, vehicle wheel, etc. Some targets may be able to exactly measure its moving distance, e.g., a vehicle equipped with speed sensors. The turning angle of the target can be obtained by using an angle measurement or compass application installed in a mobile phone or a steering wheel angle sensor installed in a vehicle. Especially, a 90° turning angle can be easily achieved through human eye estimation if the terrain allows a right-angle turn. Extensive simulations are conducted to evaluate the performance of the algorithms, investigating the effects of a set of parameters. The proposed SAL scheme not only reduces the complexity and cost of the localization system, but also provides targets with a localization opportunity in a harsh environment where only one anchor can be attached.

Location region estimation (LRE) is a key issue for many location-based applications in the Internet of Things era. This paper explores the problem of accurate LRE (ALRE) with distance distribution methods. First, in order to capture the uncertainties during the distance ranging process, a disk error model is introduced by modeling the target as a random node inside a disk region. Then, a disk error-based ranging (DEBR) approach is designed and analyzed by proving that the parameter estimation of DEBR is unbiased. Furthermore, an ALRE algorithm is developed through taking into account both DEBR and the classical multilateration method. It is proved that the estimated region obtained by ALRE is tighter than that obtained by the traditional estimation method. In addition, extensive simulations are conducted to verify the unbiased estimation of DEBR and evaluate the performance of ALRE.

The wide proliferation of drones has posed great threats to personal privacy and public security, which makes it urgent to monitor and locate intruding drones in sensitive areas. In Direction of Arrival (DOA) based localization, the estimation accuracy of DOA directly affects the localization accuracy. In this paper, we propose a novel algorithm to estimate the DOA of an intruding drone by exploiting its acoustic feature, which is mainly reflected in the strength distribution of the harmonics of the received acoustic signal. Specifically, this algorithm first estimates the harmonic frequencies of the drone's acoustic signal in frequency domain. Then, multiple signal classification is used to estimate the DOAs of all the selected harmonics. Furthermore, weighted sum of these DOA estimates will be taken as the drone's DOA estimate, where the weights are in proportional to the energy of the corresponding harmonics. The performance of the proposed algorithm is verified by both simulation and field experiments.

Privacy-preserving localization in crowdsourcing has drawn much attention recently. Under the classical nonadjacent subtraction-based localization (NSL) model, existing solutions based on homomorphic encryption techniques are of high computational and communication overheads. In this paper, an adjacent subtraction-based localization (ASL) model is first proposed. Then, an efficient privacy-preserving localization (EPPL) algorithm is developed under ASL without using any homomorphic encryption technique. In terms of the correctness, privacy, and efficiency, a comprehensive analysis is presented to investigate EPPL's performance. Furthermore, the statistical equivalence between ASL and NSL is proved through the fact that the difference between their average location estimation results converges toward zero. The lower and upper bounds of the localization error are also derived for ASL under a bounded noise model. Extensive simulations are conducted to illustrate the equivalence between ASL and NSL, and the performance of EPPL regarding the correctness, privacy, and efficiency.

Cooperative geolocation has attracted significant research interests in recent years. A large number of localization algorithms rely on the availability of statistical knowledge of measurement errors, which is often difficult to obtain in practice. Compared with the statistical knowledge of measurement errors, it can often be easier to obtain the measurement error bound. This paper investigates a localization problem assuming unknown measurement error distribution except for a bound on the error. We first formulate this localization problem as an optimization problem to minimize the worst case estimation error, which is shown to be a nonconvex optimization problem. Then, relaxation is applied to transform it into a convex one. Furthermore, we propose a distributed algorithm to solve the problem, which will converge in a few iterations. Simulation results show that the proposed algorithms are more robust to large measurement errors than existing algorithms in the literature. Geometrical analysis providing additional insights is also provided.

Non-line-of-sight (NLOS) propagation, which widely exists in wireless systems, will degrade the performance of wireless positioning system if it is not taken into consideration in the localization algorithm design. The 3rd Generation Partnership Project (3GPP) suggests that the probabilities of line-of-sight (LOS) and NLOS are related to the distance between the receiver and the transmitter. In this paper, we propose a Maximum Likelihood Estimator (MLE) for localization, which incorporates the distance dependent LOS/NLOS probabilities. Then, the position error bound is derived using Cramer-Rao Lower Bound (CRLB). Through numerical analysis, the impact of NLOS propagation on the position error bound is evaluated. The performance of our proposed algorithm is verified by real world experimental data.

We investigate a localization problem using time-difference-of-arrival measurements with unknown and bounded measurement errors. Different from most existing algorithms, we consider the minimization of the worst-case position estimation error to improve the robustness of the algorithm. The localization problem is formulated as a nonconvex optimization problem. We adopt semidefinite relaxation to relax the original problem into a convex optimization problem, which can be solved using existing semidefinite program solvers. Simulation results show that our proposed algorithm has lower worst-case position estimation error than other existing algorithms.

The size of target will induce a degradation of tracking performance, which has been neglected for simplicity in most previous studies. In multiple target tracking, occlusions will be caused by target size effect, one target can become a moving obstacle blocking the direct channel between the anchor and another target. In this paper, the data association problem in multiple target tracking is investigated. To reduce the computational complexity of traditional Joint Probabilistic Data Association (JPDA) algorithm, a modified JPDA algorithm is proposed to execute data association in multiple target tracking by utilizing the information of occlusion conditions, which is identified by a three-step algorithm. Simulation results show that the proposed algorithm is with good tracking performance and low computational complexity.

Knowing channel sight condition is important as it has a great impact on localization performance. In this paper, a RSS-based localization algorithm, which jointly takes into consideration the effect of channel sight conditions, is investigated. In our approach, the channel sight conditions experience by a moving target to all sensors is modeled as a hidden Markov model (HMM), with the quantized measured RSSs as its observation. The parameters of HMM are obtained by an off-line training assuming that the LOS/NLOS can be identified during the training phase. With the HMM matrices, a forward-only algorithm can be utilized for real time sight conditions identification. The target is localized by extended Kalman Filter (EKF) by suitably combining with the sight conditions. Simulation results show that our proposed localization strategy can provide good identification to channel sight conditions, hence results in a better localization estimation.

In this paper, the localization accuracy of range-only sensors is investigated. Being different from most previous studies, we consider a more general measurement model with both additive and multiplicative noise other than with only additive noise. The main contributions lie in twofold. First, a CRLB(Cramer-Rao Lower Bound)-based metric is proposed to evaluate the localization accuracy. In addition, the analytical relationships between target-sensor distance and localization accuracy, and between noise and localization accuracy are derived. Second, numerical analysis is executed to evaluate the localization accuracy for three important regular patterns of sensor deployment, i.e., triangle, square and hexagon. Two aspects have been examined and discussed, including (a) the geometric distribution of localization accuracy, (b) the average localization accuracy. Both theoretical and numerical results show that the multiplicative noise will influence significantly the localization accuracy. This study also provides important guidelines for optimal sensor deployment.