In this article, a survey of techniques for tracking multiple targets in distributed sensor networks is provided and introduce some recent developments. The single target reviewed. resource management issues can be readily extended to MTT. MTT problem also briefly reviewed describe the traditional approaches centralized systems. Then focus on resource-constrained present two distinct example methods demonstrating how limited resources utilized applications. Finally, most important remaining...

In a sensor network, data routing is tightly coupled to the needs of sensing task, and hence application semantics. This paper introduces novel idea information-directed routing, in which formulated as joint optimization transport information aggregation. The objective minimize communication cost, while maximizing gain, differing from considerations for more general ad hoc networks. uses concrete problem locating tracking possibly moving signal sources an example generation process,...

Conventional single LiDAR systems are inherently constrained by their limited field of view (FoV), leading to blind spots and incomplete environmental awareness, particularly on robotic platforms with strict payload limitations. Integrating a motorized offers practical solution significantly expanding the sensor's FoV enabling adaptive panoramic 3D sensing. However, high-frequency vibrations quadruped robot introduce calibration challenges, causing variations in LiDAR-motor transformation...

Most distributed algorithms for robot coordination require relative location information, but how to obtain locations in a manner is still primary problem address multi-robot applications. In order the between robots, no matter whether they are motion or stationary situation, we design rotating ultra-wideband tag provide persistency of excitation condition and two estimation estimate manner. Moreover, our approach relies only on on-board sensors requires one per robot, eliminating need any...

Offline reinforcement learning (RL) aims to learn a policy using only pre-collected and fixed data. Although avoiding the time-consuming online interactions in RL, it poses challenges for out-of-distribution (OOD) state actions often suffers from data inefficiency training. Despite many efforts being devoted addressing OOD actions, latter (data inefficiency) receives little attention offline RL. To address this, this paper proposes cross-domain which assumes incorporate additional...

The ability to characterize sensing quality is central the design and deployment of practical distributed sensor networks. This paper introduces concept a field defining, for each point in physical space phenomenon interest, measure how well network can sense at that point. Using target localization tracking as examples, derives an upper bound this goodness measure, using Cramer-Rao models observation layout. It then evaluates validity statistical used by family estimators. Simulation...

Offline reinforcement learning algorithms promise to be applicable in settings where a fixed dataset is available and no new experience can acquired. However, such formulation inevitably offline-data-hungry and, practice, collecting large offline for one specific task over environment also costly laborious. In this paper, we thus 1) formulate the dynamics adaptation by using (source) data collected from another relax requirement extensive (target) data, 2) characterize shift problem which...

There has been a lot of work on localization for sensor networks. Most schemes assume an already deployed network. Very little research done active deployment with mobile robots. In this paper, we present distributed sequential process which is tightly integrated robotic deployment. Being location-aware, sensors can be placed in advantageous locations to avoid big errors. turn, the accurate result helps further We compare our scheme five existing algorithms, both through simulations and...

Time series prediction plays a key role in wide applications and has been investigated for couple of decades. Nevertheless, most the prior works fail to identify effective frequency components time before passing through prediction, which induces drop performance. In this paper, we propose novel predictor integrates sequence (seq2seq) model based on long short-term memory units (LSTM) with interpretable data reconstruction, where learned hidden state is taken as bridge. The reconstructor can...

In this paper, we present \textbf{C}ont\textbf{E}xtual \textbf{I}mitation \textbf{L}earning~(CEIL), a general and broadly applicable algorithm for imitation learning (IL). Inspired by the formulation of hindsight information matching, derive CEIL explicitly embedding function together with contextual policy using embeddings. To achieve expert matching objective IL, advocate optimizing variable such that it biases towards mimicking behaviors. Beyond typical from demonstrations (LfD) setting,...

In recent years, the Android operating system for mobile terminals has developed very quickly. A variety of devices which are using more than 60% in domestic market share. With number application raising fast, a information leakage, malicious chargeback, failure events occurred frequently; safety also attracts wide attention researchers. this paper, combining static analysis and dynamic analysis, we present code detection method implementation. Through statistics sensitive API functions...

To realize efficient remote human-computer interaction of robots, a robot operating system based on virtual reality and digital twin is proposed. The builds model the Unity 3D engine to establish connection with entity, assisting online programming real-time manipulation unit. uses HTC VIVE build framework. actualize mutual drive between real space space, mathematical constructed through forward inverse kinematics robot. Through combination eye-tracking-based eye movement unique controller...

Intrusion Detection Systems (IDS) are critical secu-rity mechanisms that protect against a wide variety of network threats and malicious behaviors on networks or hosts. As both Network-based IDS (NIDS) Host-based (HIDS) have been widely investigated, this paper aims to present Combined System (CIDS) integrates host data in order improve performance. Due the scarcity datasets include packet data, we novel CIDS dataset formation framework can handle log files from operating systems align...

In this paper, we formulate the task of multi-robot object collaborative lifting as a distributed optimization problem. The considered total objective function is sum private functions corresponding to each robot, which are used evaluate their location choices. To avoid large horizontal component force on system during lifting, coupled equality constraint introduced formulated feasibility constraints optimal also in paper. A novel continuous-time algorithm proposed solve for second-order...

Offline reinforcement learning (RL) algorithms can improve the decision making via stitching sub-optimal trajectories to obtain more optimal ones. This capability is a crucial factor in enabling RL learn policies that are superior behavioral policy. On other hand, Decision Transformer (DT) abstracts decision-making as sequence modeling, showcasing competitive performance on offline benchmarks, however, recent studies demonstrate DT lacks of capability, thus exploit for vital further its...

In this paper, we propose a novel approach called DIffusion-guided DIversity (DIDI) for offline behavioral generation. The goal of DIDI is to learn diverse set skills from mixture label-free data. We achieve by leveraging diffusion probabilistic models as priors guide the learning process and regularize policy. By optimizing joint objective that incorporates diversity diffusion-guided regularization, encourage emergence behaviors while maintaining similarity Experimental results in four...

The robustness to distribution changes ensures that NLP models can be successfully applied in the realistic world, especially for information extraction tasks. However, most prior evaluation benchmarks have been devoted validating pairwise matching correctness, ignoring crucial validation of robustness. In this paper, we present first benchmark simulates open real where syntactic and expressive distributions under same knowledge meaning may drift variously. We design annotate a large-scale...

The linear programming model of enterprise human resource training program is established by taking maximization the total increased output as an objective function. And optimal solutions are obtained for different assessment periods. Based on comparative analysis schemes periods, it found out that period has significant effects design program. For short-term period, scheme different. However, long-term there only one scheme, which gives maximal enterprise. It further shown scientific method...

Objective evaluation of quantitative imaging (QI) methods using measurements directly obtained from patient images is highly desirable but hindered by the non-availability gold standards. To address this issue, statistical techniques have been proposed to objectively evaluate QI without a standard. These assume that measured and true values are linearly related slope, bias, normally distributed noise term, where it assumed term between different independent. However, could be correlated...

To solve the motion planning of live working manipulator, this research proposes a hybrid data-model–driven algorithm called P-SAC algorithm. In model-driven part, to avoid obstacles and make trajectory as smooth possible, we designed model sextic polynomial used PSO optimize parameters model. The data generated by part are then passed into replay buffer pre-train agent. Meanwhile, guide manipulator in reaching target point, propose reward function design based on region guidance....

It is of significance for an agent to learn a widely applicable and general-purpose policy that can achieve diverse goals including images text descriptions. Considering such perceptually-specific goals, the frontier deep reinforcement learning research goal-conditioned without hand-crafted rewards. To this kind policy, recent works usually take as reward non-parametric distance given goal in explicit embedding space. From different viewpoint, we propose novel unsupervised approach named...