Parkinson's Disease (PD) is a progressive neurodegenerative disorder marked by motor and non-motor symptoms, which complicate both diagnosis and disease monitoring. Traditional subjective assessments, such as the Unified Parkinson's Disease Rating Scale (MDS-UPDRS) and Hoehn and Yahr scale, often fall short of capturing subtle motor variations across PD stages as well as require expertise and thus the accessibility is limited in daily life. In this study, we evaluated the feasibility of using in-phase (IP) and anti-phase (AP) heel- and toe-tapping as biomarkers to monitor PD progression objectively. Motion data were collected from 40 participants (28 patients with PD, 12 age-matched healthy controls) using a pair of smart insoles with embedded accelerometers. Our results show that the heel IP yielded the largest number of stage-discriminative features, with 96 out of 112 extracted features showing significant differences across PD stages and achieved up to 92% classification accuracy using supervised machine learning classifiers, particularly Random Forest and Neural Network. Clustering analyses (KMeans and Gaussian Mixture Model) further supported stage-specific grouping patterns, with chi-square significance p < 0.0001. These findings suggest that heel IP can serve as a sensitive and practical assessment for PD stage classification. Paired with smart insoles and a tapping game, this can be used as an assistive tool to monitor PD disease progression for at home monitoring in daily life.

Exploring the Machine Learning (ML) features and methods based on electroencephalography (EEG) and quantitative electroencephalographic (qEEG) holds the potential to early diagnosis and classification of Parkinson’s disease (PD) and can bring convenience to the decision-making process for healthcare professionals. PD is clinically heterogeneous, with high variability between patients in terms of motor and non-motor symptoms. This nature makes it extremely challenging for early diagnosis, and severity classification using traditionally questionnaire-based standards. EEG and qEEG provide objective signals with rich features that can potentially provide a translational alternative. In the past decade, a host of ML-based features and algorithms have been developed to detect and classify PD in patients using a range of EEG and qEEG datasets. Various methodologies, including graph-CNNs, deep RNNs, and hybrid models, are discussed within the context of early diagnosis and classification, each offering unique insights and performance levels. This review highlights the multifaceted nature of PD diagnosis and classification research. It addresses the challenges and limitations encountered in existing techniques, emphasizing the need for further improvements in accuracy, sensitivity, and specificity. This review provides a standardized insight into the evolving landscape of PD diagnosis and classification. It showcases the remarkable progress that has been made in utilizing EEG and qEEG data while emphasizing the ongoing quest for effective and convenient ML based methods for early diagnosis and classification.

Accurate occupancy detection remains a challenging problem due to dynamic occupancy patterns and varying environments. Traditional machine learning (ML) struggles with this variability as models typically require large datasets and frequent updates as occupancy scenarios are unlimited and continuously change over time. It is thus infeasible to train a single "universal" model for the diverse real-world scenarios given real-world computational constraints. To address these issues, a context-aware hierarchical classification framework is proposed which periodically trains multiple occupancy classifiers on subsets of data delineated by meaningful contexts. When new occupancy data arrive, its context is identified, and a corresponding pretrained classifier is selected for prediction. By focusing each model on more consistent data distributions defined by context, this approach aims to improve classification accuracy compared to baselines trained on static datasets alone. The framework also aims to eliminate the need for offline training on large datasets and frequent overthe-cloud model updates required by traditional ML approaches by performing ML-based training and inference directly on the sensor node via an Internet-of-Things (IoTs) device. The framework is evaluated via datasets collected both in an office and a residential setting, monitored by a network of synchronized low-energy electronically chopped passive infra-red (SLEEPIR) sensors. These sensors, unlike conventional passive infrared (PIR) sensors, can detect stationary occupants. Time-series features are extracted from observations and clustered to discover underlying contextual scenarios. Experimentation resulted in context scenarios which essentially represent varying levels of infrared (IR) noise in observed environment. The proposed framework achieved a 5.03% accuracy improvement over the best baseline algorithm.

Endoscopy is widely employed for diagnostic examination of the interior of organs and body cavities and numerous surgical interventions. Still, the inability to correlate individual 2D images with 3D organ morphology limits its applications, especially in intra-operative planning and navigation, disease physiology, cancer surveillance, etc. As a result, most endoscopy videos, which carry enormous data potential, are used only for real-time guidance and are discarded after collection. We present a complete method for the 3D reconstruction of inner organs that suggests image extraction techniques from endoscopic videos and a novel image preprocessing technique to reconstruct and visualize a 3D model of organs from an endoscopic video. We use advanced computer vision methods and do not require any modifications to the clinical-grade endoscopy hardware. We have also formalized an image acquisition protocol through experimentation with a calibrated test bed. We validate the accuracy and robustness of our reconstruction using a test bed with known ground truth. Our method can significantly contribute to endoscopy-based diagnostic and surgical procedures using comprehensive tissue and tumor 3D visualization.

Our recently developed synchronized low-energy electronically chopped passive infrared (SLEEPIR) sensor node enables the stationary occupancy detection capability of traditional passive infrared (PIR) sensors. A machine learning (ML) algorithm reports occupancy based on a locally collected dataset from the sensor node. Though promising, the ML algorithm’s detection accuracy depends on the diversity of the collected dataset—provided that the dataset contains a wide variety of infrared (IR) noise and occupancy patterns. Thus, it is challenging to train a universal ML model that contains all possible patterns. We propose an efficient ${K}$ -nearest neighbor (KNN) occupancy classifier that incrementally adapts to the novel data from the sensor. The proposed algorithm ensures that only the relevant noise and occupancy patterns are learned. The fact that training observations are gathered on the same sensor node where the inference is made keeps the proposed classifier accurate even with the bounded size of the dataset. A small dataset and an architecture like KNN both enable the training and inference to be executed on a resource-constrained Internet of Things (IoT) device. Thus, the proposed on-device lifelong learning (ODLL) approach eliminates the need for over-the-cloud ML model updates. The dataset was collected for two distinct floorplans over two months. Results indicate an average occupancy accuracy improvement of 20.8% compared to a statically trained long short-term memory (LSTM) model. The proposed KNN model delivers comparable detection accuracy while remaining orders of magnitude faster in terms of computational performance when compared to the LSTM-based occupancy detection algorithm.

We have previously developed a synchronized low-energy electronically chopped passive infrared (SLEEPIR) sensor node that can detect both stationary and moving occupants. In this article, we present a Bayes filter (BF)-based network-level algorithm that uses a network of SLEEPIR sensor nodes deployed at a residential apartment to estimate the occupancy of the entire apartment. The method processes the incoming observations from each of the sensor nodes via a sensor model and transforms these observations into Bayesian updates. The sensor model uses a Markov decision process (MDP) formulation to estimate the temporal bounds on the rate of occupancy flow between one occupancy state to another. The overall BF output is a probability density function (pdf) that represents the occupancy state of the entire observed space. The sensor node adjacency matrix and observation frequency are the key parameters that contribute to the sensor model design. The sensor model uses estimated transition time and probability between occupancy states to filter out observations that do not conform to the constraints set forth by the parameters. Occupancy is established through a thresholding function applied to the output pdf of the BF. A dataset was collected at a residential unit over a period of one month using the SLEEPIR sensor system. Results indicate an average 23.68% occupancy accuracy improvement when compared to the accuracy state delivered by individual SLEEPIR nodes. Results also indicate a 7.74% occupancy accuracy improvement when compared to the accuracy state determined by an already proposed particle filter (PF)-based occupancy estimation algorithm.

In the past decade, different sensing mechanisms and algorithms have been developed to detect or estimate indoor occupancy. One of the most recent advancements is using networked sensor nodes to create a more comprehensive occupancy detection system where multiple sensors can identify human presence within more expansive areas while delivering enhanced accuracy compared to a system that relies on stand-alone sensor nodes. The present work reviews the studies from 2012 to 2022 that use networked sensor nodes to detect indoor occupancy, focusing on PIR-based sensors. Methods are compared based on pivotal ADPs that play a significant role in selecting an occupancy detection system for applications such as Health and Safety or occupant comfort. These parameters include accuracy, information requirement, maximum sensor failure and minimum observation rate, and feasible detection area. We briefly describe the overview of occupancy detection criteria used by each study and introduce a metric called "sensor node deployment density" through our analysis. This metric captures the strength of network-level data filtering and fusion algorithms found in the literature. It is hinged on the fact that a robust occupancy estimation algorithm requires a minimal number of nodes to estimate occupancy. This review only focuses on the occupancy estimation models for networked sensor nodes. It thus provides a standardized insight into networked nodes' occupancy sensing pipelines, which employ data fusion strategies, network-level machine learning algorithms, and occupancy estimation algorithms. This review thus helps determine the suitability of the reviewed methods to a standard set of application areas by analyzing their gaps.

We recently developed a synchronized low energy electronically chopped passive infrared (SLEEPIR) sensor that can detect both stationary and moving occupants by incorporating a liquid crystal (LC) shutter with a traditional passive infrared (PIR) sensor. However, its detection accuracy is still largely impacted by environmental infrared noises. In this paper, we present a Particle Filter (PF) based system-level algorithm that employs a network of SLEEPIR sensors which are installed at different points of interest within an indoor space. The method interprets the incoming observations from the field of view (FOV) of each sensor via the likelihood function to update the state of the PF. The PF output is a probability density function (pdf) that represents the occupancy state of the entire observed space. The sensor location, observation cone, range, observation frequency and historic inter-sensor correlation are the key parameters that contribute to the likelihood function design. Since the method utilizes the historic correlation among sensors, the pairs of correlating sensors often perform self-correction whenever a faulty observation is encountered due to either sensor limitations or due to environmental noise. Occupancy is established through a thresholding function applied to the output pdf of the PF. A lab-based dataset was collected over a period of 360 hours using the SLEEPIR sensor system. Results indicate an average 8.25% occupancy accuracy improvement when compared to the accuracy state delivered by individual SLEEPIR nodes.

This paper presents an energy-efficient classification framework that performs human activity recognition (HAR). Typically, HAR classification tasks require a computational platform that includes a processor and memory along with sensors and their interfaces, all of which consume significant power. The presented framework employs microelectromechanical systems (MEMS) based Continuous Time Recurrent Neural Network (CTRNN) to perform HAR tasks very efficiently. In a real physical implementation, we show that the MEMS-CTRNN nodes can perform computing while consuming power on a nano-watts scale compared to the micro-watts state-of-the-art hardware. We also confirm that this huge power reduction doesn't come at the expense of reduced performance by evaluating its accuracy to classify the highly cited human activity recognition dataset (HAPT). Our simulation results show that the HAR framework that consists of a training module, and a network of MEMS-based CTRNN nodes, provides HAR classification accuracy for the HAPT that is comparable to traditional CTRNN and other Recurrent Neural Network (RNN) implantations. For example, we show that the MEMS-based CTRNN model average accuracy for the worst-case scenario of not using pre-processing techniques, such as quantization, to classify 5 different activities is 77.94% compared to 78.48% using the traditional CTRNN.

The goal of this paper is to provide a novel computing approach that can be used to reduce the power consumption, size, and cost of wearable electronics. To achieve this goal, the use of microelectromechanical systems (MEMS) sensors for simultaneous sensing and computing is introduced. Specifically, by enabling sensing and computing locally at the MEMS sensor node and utilizing the usually unwanted pull in/out hysteresis, we may eliminate the need for cloud computing and reduce the use of analog-to-digital converters, sampling circuits, and digital processors. As a proof of concept, we show that a simulation model of a network of three commercially available MEMS accelerometers can classify a train of square and triangular acceleration signals inherently using pull-in and release hysteresis. Furthermore, we develop and fabricate a network with finger arrays of parallel plate actuators to facilitate coupling between MEMS devices in the network using actuating assemblies and biasing assemblies, thus bypassing the previously reported coupling challenge in MEMS neural networks.

The size and power limitations in small electronic systems such as wearable devices limit their potential. Significant energy is lost utilizing current computational schemes in processes such as analog-to-digital conversion and wireless communication for cloud computing. Edge computing, where information is processed near the data sources, was shown to significantly enhance the performance of computational systems and reduce their power consumption. In this work, we push computation directly into the sensory node by presenting the use of an array of electrostatic Microelectromechanical systems (MEMS) sensors to perform colocalized sensing-and-computing. The MEMS network is operated around the pull-in regime to access the instability jump and the hysteresis available in this regime. Within this regime, the MEMS network is capable of emulating the response of the continuous-time recurrent neural network (CTRNN) computational scheme. The network is shown to be successful at classifying a quasi-static input acceleration waveform into square or triangle signals in the absence of digital processors. Our results show that the MEMS may be a viable solution for edge computing implementation without the need for digital electronics or micro-processors. Moreover, our results can be used as a basis for the development of new types of specialized MEMS sensors (ex: gesture recognition sensors).

Recently, stereovision has appeared in robotics as a source of information for real-time mapping and path planning. In this paper, an intelligent motion system for mobile robots is designed and implemented using stereovision. The proposed system uses stereovision as a primary method for sensing the environment, and the system is able to navigate intelligently in an indoor environment with varying degrees of obstacle complexity. It creates noiseless and high-confidence 3D point clouds and uses these point clouds as an input for the mapping and path-planning modules. The proposed system was built by developing, enhancing, and integrating various techniques, modules and algorithms. The Stereovision-based Path-planning module is the integration of three main enhanced techniques: (1) the multi-baseline multi-view stereovision filter (MMSVF), (2) accurate floor detection and segmentation (AFDS), and (3) the intelligent gazing module (IGM). This Stereovision-based Path planning (MMSVF, IGM, and AFDS) was integrated with the Fuzzy Logic Motion Controller (FLMC). All techniques, modules and algorithms are implemented using a multi-threaded and client–server-based architecture. To prove the viability and robustness of our proposed system, we have integrated all components of the system into a fully functional mobile robot navigation system. We compared the performance of the main modules with that of similar modules in the literatures, and showed that our modules had better performance. Testing the whole system is more important than just testing each module individually. To the best of our knowledge, the literatures lack such testing. Hence, in this paper we present the performance of our complete integrated system in different environments using different parameters and different architectures.

The field of robotics has apparently appeared to replace jobs that humans usually dislike doing, jobs that have low tolerances for faults or that are dangerous for humans. For these reasons, mobile robots have been used in many applications, such as gas detection, security patrolling, and search and rescue. Nowadays, many gases that are dangerous for humans may pollute the environment. Gases can leak from many causes, such as mining disasters, war, and manufacturing disasters. In this paper, we introduce a new mobile robotics system, which uses a mobile robot for navigation within hazardous environments to detect, localize, and create concentration maps of various gases that are deemed hazardous in household and industrial environments. The generated gas maps can be used for many purposes; for example, they might be used by humans to avoid the areas with detected gas or by a rescue team during rescue operations.

Mobile robots provide automated solutions for a range of tasks in industrial settings including but not limited to inspection. Our interest is automated inspection tasks including gas leakage detection in natural gas processing facilities such as those in Qatar. Using autonomous mobile robot solutions remove humans from potentially hazardous environments, eliminate potential human errors from fatigue, and provide data logging solutions for visualization and off-line post-processing. A core requirement for a mobile robot to perform any meaningful inspection task is to localize itself within the operating environment. We are developing a visual Simultaneous Localization And Mapping (SLAM) system for this purpose. Visual SLAM systems enable a robot to localize within an environment while simultaneously building a metric 3D map using only imagery from an on-board camera head. Vision has many advantages over alternate sensors used for localization and mapping. It requires minimal power compared to Lidar sensors, is relatively inexpensive compared to Inertial Navigation Systems (INS), and can operate in GPS denied environments. There is extensive work related to visual SLAM with most systems using either a perspective stereo camera head or a wide-angle of view monocular camera. Stereo cameras enable Euclidean 3D reconstruction from a single stereo pair and provide metric pose estimates. However, the narrow angle of view can limit pose estimation accuracy as visual features can typically be 'tracked' only across a small number of frames. Moreover, the limited angle of view presents challenges for place recognition whereby previously visited locations can be detected and loop closure performed to correct for long-range integrated position estimate inaccuracies. In contrast, wide-angle of view monocular cameras (e.g. fisheye and catadioptric) trade spatial resolution for an increased angle of view. This increased angle can enables visual scene points to be tracked over many frames and can improve rotational pose estimates. The increased angle of view can also improve visual place recognition performance as the same areas of a scene can be imaged under much larger changes in position and orientation. The primary disadvantage of a monocular wide-angle visual SLAM system is a scale ambiguity in the translational component of pose/position estimates. The visual SLAM system being developed in this work uses a combined stereo and wide-angle fisheye camera system with the aim of exploiting the advantages of each. For this we have combined visual feature tracks from both the stereo and fisheye camera within a single non-linear least-squares Sparse Bundle Adjustment (SBA) framework for localization. Initial experiments using large scale image datasets (approximately 10 kilometers in length) collected within Education City have been used to evaluate improvements in localization accuracy using the combined system. Additionally, we have demonstrated performance improvements in visual place recognition using our existing Hidden Markov Model (HMM) based place recognition algorithm.

Real world indoor environments are rich in planar surfaces. Floor detection or ground-plane detection is a crucial requirement for a robotic navigation task. Despite frequent successes in this area, problems with detection of navigable floor with multiple planar and non-planar slopes at multiple heights still exist. For robust and safe navigation, such small variations such as floor joins, carpet deformities, raised steps and floor gradients need to be detected and robot path and kinodynamics plan must be adjusted accordingly. The authors suggest a recursive RANSAC segmentation based algorithm that estimates the dominant and sub-dominant plane models for all the navigable planes within a detected floor or a ground plane. The algorithm also divides the input point clouds intelligently into multiple regions of interest for both efficiency and accuracy enhancement. The recursive estimation approach for determining plane parameters helps to detect multiple planes within each region. Among other benefits of this approach, reduction of search space size for the estimation of plane parameters stands out to be the most striking result of this work. This region wise plane estimation approach also helps to reduce the computational load by selectively dropping less significant floor sections from estimation process. The floor estimation technique coupled with sensor response functions for two different point cloud generators further investigates into the robustness of the method when deployed on two distinct sensors i.e. RGB+D sensor and a stereo vision camera. In our experiments we segment navigable floor planes in real-time for a slowly moving sensor. The location and geometrical parameters of the floor planes are updated in a global coordinate system whenever a change their location is detected. The planes are associated to a grid map which serves as a path-planning reference to a mobile robot used in our experiments. The results of floor detection and the precision of floor anomaly detection are compared sensor-wise and with the ground truth defined by obstacle heights and configuration.

Studies have shown that computations of visual kinematics relations are highly complicated and do require a considerable amount of time. This is due to dependent on Jacobians and massive inter-related relations. This hinders even complicated visual servo algorithm for real-time applications. In this respect, the proposed methodology is based on approximating the highly nonlinear relations and epipolar geometry relating changes of object visual features to changes in joint space of a robotics arm system, which are usually expressed in terms of kinematics relations, in addition to time-dependent Jacobian matrix. Artificial Neural system have been proposed for that purpose. A supervised learning artificial neutral network have been employed for learning the visual nonlinear kinematics relations. For validation, the concept have been applied to the well known Rives visual servo algorithm [1], with Two Scenes Epipolar Geometry. Results have shown that, highly accurate visual serving was achieved, while considerable amount of time has been reduced with the proposed methodology

A novel method is proposed that adapts a previously proposed LADAR based pedestrian detection and tracking technique by introducing a stereo-vision based segmentation technique for the purpose of pedestrian detection and tracking. The proposed method detects the harmonic motions of limbs and body during a typical human walk and temporally propagates the position, stride, direction and phase using a particle filter. The particle-filter uses a human limb-motion model and is able to track the walking pedestrians in a heavily occluded environment. Potential 3D point clusters belonging to arms and feet are extracted employing an adapted version of RANSAC based segmentation algorithm. A Fourier-transform based periodogram confirms the periodicity for each point-cluster representing limbs. Since RGB or intensity data from the stereo-vision input is ignored and the proposed method completely relies upon 3D data produced by the stereo-vision sensor, reliable illumination invariant pedestrian detection and tracking results are achieved using Daimler-Stereo-Pedestrian-Detection-Dataset. Further lab experiments also confirm the viability of the method within the indoor environment.

This manuscript looks into details into the hardware implementations of an active stereo vision path planning navigation for a mobile robotics system. This mobile robotics system has been designated as the (KSU-IMR), as it a test-bed for research within the area of mobile robotics system navigation through the use of an active stereo vision for planning a path within an unstructured environment at King Saud University in KSA. The system is making use of fast image processing tools, as in addition to fast algorithms for processing the dynamic images. The approach followed here is to employ the Epipolar Geometry to compute the 3-D details of the environment where the mobile is moving. The algorithm mixes stereo vision, occupancy grid mapping, and potential field path planning procedures to form a robust and consistent robotic system for mapping and navigation.

Real world indoor environments are rich in planar surfaces. Floor detection or ground-plane detection is a crucial requirement for a robotic navigation task. Despite frequent successes in this area, problems with detection of navigable floor with multiple planar and non-planar slopes at multiple heights still exist. For robust and safe navigation, such small variations such as floor joins, carpet deformities, raised steps and floor gradients need to be detected and robot path and kinodynamics plan must be adjusted accordingly. The authors suggest a recursive RANSAC segmentation based algorithm that estimates the dominant and sub-dominant plane models for all the navigable planes within a detected floor or a ground plane. The algorithm also divides the input point clouds intelligently into multiple regions of interest for both efficiency and accuracy enhancement. The recursive estimation approach for determining plane parameters helps to detect multiple planes within each region. Among other benefits of this approach, reduction of search space size for the estimation of plane parameters stands out to be the most striking result of this work. This region wise plane estimation approach also helps to reduce the computational load by selectively dropping less significant floor sections from estimation process. The floor estimation technique coupled with sensor response functions for two different point cloud generators further investigates into the robustness of the method when deployed on two distinct sensors i.e. RGB+D sensor and a stereo vision camera. In our experiments we segment navigable floor planes in realtime for a slowly moving sensor. The location and geometrical parameters of the floor planes are updated in a global coordinate systemwhenever a change their location is detected. The planes are associated to a grid map which serves as a path-planning reference to a mobile robot used in our experiments. The results of floor detection and the precision of floor anomaly detection are compared sensorwise and with the ground truth defined by obstacle heights and configuration.

D* based navigation algorithms provide robust and real-time means of achieving path planning in dynamic environments. Author of this paper introduces a notion of predictable time-based obstacles. The algorithm proposed in the paper defines a centralized obstacle-map that is shared among multiple agents (robots) performing path planning. Each robot plans its path individually on an obstacle-map using a slightly modified version of D* Lite and then shares an updated version of the map, which includes its planned path as a new obstacle, with its peers. The planned paths appear as temporary time-based obstacles to peer robots. Planned paths are divided into discrete temporal sections so as to help peer robots optimize paths temporally. The proposed algorithm also presents a priority measure which helps us decide the optimized sequence of individual path-planning order followed by cooperating robots. Since the implemented algorithm is tested in simulation using Mobile robot Programming Toolkit, the Real-time performance analysis is done to confirm the real-time execution time of the proposed algorithm.