Diluka Moratuwage

Human–robot interaction (HRI) and navigation in crowded indoor spaces require robots to reason not only about where humans may be but also about whether uncertain human motion conflicts with the robot’s intended path. Existing context-aware particle filtering (CA-PF) provides a useful belief-estimation backbone by maintaining multimodal human-state uncertainty under occlusion, missed detections, and close interaction. However, a navigation layer must further convert this belief into path-aware risk and determine when replanning is necessary. This paper extends the CA-PF framework with three navigation modules: a path-conflict risk module that evaluates weighted human particles against the committed robot path, a directional interaction-aware risk grid that combines particle occupancy with human motion direction and conflict likelihood, and a belief-aware event-triggered replanning strategy that updates the path when conflict, directional path risk, or risk growth exceeds safety thresholds. Evaluation on four sequences from the Bi3 dataset demonstrates a mean human-tracking error of 0.237 m and an RMSE of 0.308 m. In the ablation study, the complete framework reduced mean path-conflict risk by 25.2% and mean directional path risk by 42.35% relative to the configuration without event-triggered replanning, averaged over 12 matched trials. These results demonstrate that propagating particle-level human beliefs into path-conflict reasoning, directional risk-aware planning, and selective replanning can improve the handling of interaction risk in dynamic human environments.

Industrial quality control is a critical component of modern manufacturing, as defects can lead to significant economic losses and safety risks. Traditional inspection methods, largely reliant on human operators or rule-based systems, often suffer from inconsistency, limited scalability, and reduced accuracy in complex environments. Recent advances in artificial intelligence (AI), particularly in deep learning and computer vision, have enabled automated defect detection and classification with unprecedented performance. This paper provides a comprehensive review of AI-based image processing techniques for industrial quality control, covering classification, detection, and segmentation approaches. Key applications across manufacturing sectors are discussed, alongside current challenges such as data scarcity, real-time implementation, and model generalisation. Furthermore, this paper explores emerging trends toward autonomous inspection systems, integrating real-time analytics, edge computing, and intelligent decision making. The insights presented aim to guide future research toward robust, scalable, and fully automated quality control solutions in smart manufacturing environments.

Electromyography (EMG) has become essential in biomedical engineering, rehabilitation, and human–machine interfacing due to its ability to capture neuromuscular activation for control, monitoring, and diagnosis. Recent advances in sensing hardware, high-density and flexible electrodes, and embedded acquisition modules combined with modern signal processing and machine learning have significantly enhanced the robustness and applicability of EMG-based systems. This review provides an integrated overview of EMG generation, acquisition standards, and preprocessing techniques, including adaptive filtering, wavelet denoising, and empirical mode decomposition. Feature extraction methods across the time, frequency, time–frequency, and nonlinear domains are compared with respect to computational efficiency and suitability for real-time systems. The review synthesizes classical and contemporary pattern-recognition approaches, from statistical classifiers to deep architectures such as CNNs, RNNs, hybrid CNN–RNN models, transformer-based networks, and graph neural networks. Key challenges, including signal non-stationarity, electrode displacement, muscle fatigue, and poor cross-user or cross-session generalization, are examined alongside emerging strategies such as transfer learning, domain adaptation, and multimodal fusion with IMU or FMG signals. Finally, the paper surveys rapidly growing EMG applications in prosthetics, rehabilitation robotics, human–machine interfaces, clinical diagnostics, and sports analytics. The review highlights ongoing limitations and outlines future pathways toward robust, adaptive, and deployable EMG-driven intelligent systems.

Electromyography (EMG) has emerged as a vital tool in the development of wearable robotic exoskeletons, enabling intuitive and responsive control by capturing neuromuscular signals. This review presents a comprehensive analysis of the EMG signal processing pipeline tailored to exoskeleton applications, spanning signal acquisition, noise mitigation, data preprocessing, feature extraction, and control strategies. Various EMG acquisition methods, including surface, intramuscular, and high-density surface EMG, are evaluated for their applicability in real-time control. The review addresses prevalent signal quality challenges, such as motion artifacts, power-line interference, and crosstalk. It also highlights both traditional filtering techniques and advanced methods, such as wavelet transforms, empirical mode decomposition, and adaptive filtering. Feature extraction techniques are explored to support pattern recognition and motion classification. Machine learning approaches are examined for their roles in pattern recognition-based and hybrid control architectures. This article emphasizes muscle synergy analysis and adaptive control algorithms to enhance personalization and fatigue compensation, followed by the benefits of multimodal sensing and edge computing in addressing the limitations of EMG-only systems. By focusing on EMG-driven strategies through signal processing, machine learning, and sensor fusion innovations, this review bridges gaps in human–machine interaction, offering insights into improving the precision, adaptability, and robustness of next generation exoskeletons.

Generalized Labeled Multi-Bernoulli (GLMB) densities arise in a host of multi-object system applications analogous to Gaussians in single-object filtering. However, computing the GLMB filtering density requires solving NP-hard problems. To alleviate this computational bottleneck, we develop a linear complexity Gibbs sampling framework for GLMB density computation. Specifically, we propose a tempered Gibbs sampler that exploits the structure of the GLMB filtering density to achieve an $\mathcal {O}(T(P+M))$ complexity, where $T$ is the number of iterations of the algorithm, $P$ and $M$ are the number hypothesized objects and measurements. This innovation enables the GLMB filter implementation to be reduced from an $\mathcal {O}(TP^{2}M)$ complexity to $\mathcal {O}(T(P+M+\log T)+PM)$ . Moreover, the proposed framework provides the flexibility for trade-offs between tracking performance and computational load. Convergence of the proposed Gibbs sampler is established, and numerical studies are presented to validate the proposed GLMB filter implementation.

If computational tractability were not an issue, multi-object estimation should integrate all measurements from multiple sensors across multiple scans. In this article, we propose an efficient numerical solution to the multi-scan multi-sensor multi-object estimation problem by computing the (labeled) multi-sensor multi-object posterior density. Minimizing the $L_{1}$ -norm error from the exact posterior density requires solving large-scale multi-dimensional assignment problems that are NP-hard. An efficient multi-dimensional assignment algorithm is developed based on Gibbs sampling, together with convergence analysis. The resulting multi-scan multi-sensor multi-object estimation algorithm can be applied either offline in one batch or recursively. The efficacy of the algorithm is demonstrated using numerical experiments with a simulated dataset.

Multi-target state estimation refers to estimating the number of targets and their trajectories in a surveillance area using measurements contaminated with noise and clutter. In the Bayesian paradigm, the most common approach to multi-target estimation is by recursively propagating the multi-target filtering density, updating it with current measurements set at each timestep. In comparison, multi-target smoothing uses all measurements up to current timestep and recursively propagates the entire history of multi-target state using the multi-target posterior density. The recent Generalized Labeled Multi-Bernoulli (GLMB) smoother is an analytic recursion that propagate the labeled multi-object posterior by recursively updating labels to measurement association maps from the beginning to current timestep. In this paper, we propose a moving window based solution for multi-target tracking using the GLMB smoother, so that only those association maps in a window (consisting of latest maps) get updated, resulting in an efficient approximate solution suitable for practical implementations.

Meeting or conference assistance is a popular application that typically requires compact configurations of co-located audio and visual sensors. This paper proposes a novel solution for online separation of an unknown and time-varying number of moving sources using only a single microphone array co-located with a single visual device. The approach exploits the complementary nature of simultaneous audio and visual measurements, accomplished by a model-centric 3-stage process of detection, tracking, and (spatial) filtering, which performs separation in a block-wise or recursive fashion. Fusing the measurements requires solving the multi-modal space-time permutation problem, since the audio and visual measurements reside in different observation spaces, but also are unidentified or unlabeled (with respect to the unknown and time-varying number of sources), and are subject to noise, extraneous measurements and missing measurements. A labeled random finite set tracking filter is applied to resolve the permutation problem and recursively estimate the source identities and trajectories. A time-varying set of generalized side-lobe cancellers is constructed based on the tracking estimates to perform online separation. Evaluations are undertaken with live human speakers.