In this work, we investigate the problem of level curve tracking in unknown scalar fields using a limited number of mobile robots. We design and implement a long short term memory (LSTM) enabled control strategy for a mobile sensor network to detect and track desired level curves. Based on the existing work of cooperative Kalman filter, we design an LSTM-enhanced Kalman filter that utilizes the sensor measurements and a sequence of past fields and gradients to estimate the current field value and gradient. We also design an LSTM model to estimate the Hessian of the field. The LSTM enabled strategy has some benefits such as it can be trained offline on a collection of level curves in known fields prior to deployment, where the trained model will enable the mobile sensor network to track level curves in unknown fields for various applications. Another benefit is that we can train using larger resources to get more accurate models, while utilizing a limited number of resources when the mobile sensor network is deployed in production. Simulation results show that this LSTM enabled control strategy successfully tracks the level curve using a mobile multi-robot sensor network.

Smart farming experiential learning and research endeavors have been ongoing at the University of Maryland Eastern Shore (UMES) for the past several years. Recent field implementation of contemporary technologies for variable rate fertilizer application based on multispectral drone imagery; deployment of wireless solar powered soil moisture sensor network on a field with subsurface drip and fertigation capability; and development of a sustainable platform integrated with a Cartesian robotic device powered by solar and wind energy that can seed, weed, irrigate, and capture time-lapse photography while servicing a small raised bed for specialty crops and vegetables will be described in this paper. Results from the initial phase of implementation efforts and future goals will also be highlighted.

With enormous data being generated every day from countless applications and sensor networks, the need for efficient computing devices to process and make use of this data continues to grow and is projected to increase in the future. The idea of analog computing has been redeemed recently and is poised as the future computing paradigm in these applications due to its powerful computing power and lower energy consumption. This work presents a mechanism where MEMS devices are coupled together resulting in dynamic behaviors that qualitatively resemble that of the biologically-inspired Continuous-Time Recurrent Neural Networks (CTRNNs). Moreover, interesting oscillations behaviors and limit cycles attributed to various weight manipulations and excitation input levels have been observed.

Towers, roller batteries, propelling cables and vehicles are the substructures of ropeway transportation systems. High-fidelity modeling of their dynamical interactions together with a reliable identification is a key step towards the prediction of the system response under various transit conditions as well as to investigate design optimization strategies. In this work, a nonlinear mechanical model for the dynamical description of cablecar ski lift systems is discussed. The investigation is focused on the modal features and the forced dynamic response caused by the vehicles transit across the so-called compression towers. The model is validated according to experimental data acquired via a custom-design sensor network. The Enhanced Frequency Domain Decomposition (EFDD) method is employed to identify the frequencies and damping ratios.

Cyber-physical systems enable new digital ecologies in industrial and workplace lifelong learning. This paper reports on early efforts in delivering a virtual environment and system for vocational education and training (VET), in particular targeting the needs of craft and trade skills. The domain of stone masonry is presented herein, where its underpinning activities are learning through virtual environments, simulation and role play. The challenges are not only the synchronicity between physical and software components but also the in-game mechanics that incorporate building blocks of effective training and skills development strategies. A prototype body-area sensor network in a cyber-physical game environment demonstrates the interaction between virtual objects and the player-learner.

The aging population, prevalence of chronic diseases, and depression are some of the major challenges of current healthcare. The smart home monitoring provides a solution of human-centric out-of-hospital care and calls for more attention to improve the comfort of patients. Among home technologies, electric lighting system is ubiquitous and affecting many aspects of human physiology, including not only visual system but also non-image-forming system. The exposure to light can influence various states of human and lead to effective comforting. This paper presents a human-centric smart lighting system controlled by human state using ubiquitous physiological sensor networks. The proposed system, acting as an intelligent home monitoring component, can naturalistically detect human’s state such as intensiveness, sleepiness, and depression. Vital signals including ECG and its secondary parameters can be acquired by non-intrusive physiological sensors and wirelessly transmitted to a control system. The color of LED can be controlled by the detected human state and adjusted to emotional comfort. The results show that different colors of light can influence human’s state.

The challenge to enhance the vehicle driving and handling with a state estimation and prediction system is presented by fusing a primary real time multibody vehicle model capable of providing a good indication of vehicle stability and control, and a secondary model able to estimate the vehicle state from vehicle real and virtual sensors to correct the indications of the primary model. A mathematical algorithm combines these two models in the drive control system improving the behavior of the active systems of the vehicle. A Multibody vehicle model has been used to achieve a high fidelity simulation of vehicle dynamics. The selected software is LMS.Virtual.Lab Motion with Real-Time Solver which complements the AMESim Real-Time Solver to handle complex real-time 3D-1D mechatronic systems without any simplified conceptual models. A Sensor Signal Processing Model has been developed to estimate the vehicle states and calculating tire-road contact forces and vehicle sideslip angle. The methodological approach uses the equations of motion of the chassis applying the fundamental principles of classical physics: Newtonian method and Euler angles. The control logic is based on the continuous updating of the preview multibody vehicle model by the controller sensors information network, which makes the model forecast behavior closer to the real one and improve comfort and linearity of the vehicle response. The driver inputs (throttle, steer angle and torque, brake, gear) are the same for the MBS real time model and for the real vehicle. A first training logic updates the MBS model based on the real vehicle behavior calculated by the sensor network, where the logic has to update in the MBS model just the factors depending on the vehicle itself (for example car weight, tire temperature, shock absorber damping forces, tires characteristics) and to understand and keep into account different environment variation (wet / dry surface). If the real vehicle is equipped with active control systems to improve handling and stability, as active camber control, drive by wire, ESP, Body movement active controls, the real time multibody model will interact with the models 1D or 3D of these vehicle dynamics controls and will improve their performance with a very high accuracy prediction of their influence on vehicle dynamic response. In conclusion with the help of the preview multibody vehicle model the drive control logic will increase the performance and drive ability of the vehicle with smart logic interacting with all the active systems.

The harvesting of ambient vibrations for powering wireless electronic components has been heavily researched over the last decade. As long as sufficient vibrational energy is readily available in the neighborhood of small electronic devices, it is possible to achieve mechanical-to-electrical energy conversion by means of a proper transduction mechanism and thereby enable self-powered wireless electronic systems. An alternative scenario is the case in which the wireless electronic component has little or no vibrational energy available in its environment, yet wireless charging of its battery is still of great interest. Examples to this scenario range from medical implants to underwater sensor networks. The goal in such systems is to charge the electronic component by contactless transfer of energy without relying on mechanical or electrical connection to the wireless component. The commonly used transduction methods of wireless energy transfer are known to be inductive, capacitive, far-field electromagnetic, and optical coupling. Acoustic energy transfer based on the propagation of energy at ultrasonic frequencies is a recently explored alternative. As this field receives growing attention, it is required to develop fully coupled modeling framework to quantify the energy transfer characteristics with a focus on the transmitter, receiver, fluid medium, as well as geometric and material parameters. In this paper, we present multiphysics modeling of contactless ultrasonic energy transfer for wireless electronic components submerged in liquid medium. The source is a pulsating sphere and the receiver is a piezoelectric transducer operating in the 33-mode of piezoelectricity with fundamental underwater resonance frequency above the audible frequency range. The goal is to quantify the electrical power transmitted from the source to the electrical load across the electrodes of the receiver in terms of the source strength for a given distance between the transmitter and the receiver. Both analytical and finite-element models are developed for the resulting acoustic-piezoelectric structure interaction problem. Fixed-free and free-free mechanical boundary conditions are considered for the fluid loaded receiver to validate the analytical model against multiphysics finite-element simulations. For the electrical boundary conditions across the electrodes, resistive and resistive-inductive loads are considered. Specifically broadband power transfer is achieved by optimal resistive-inductive load tuning for performance enhancement. Effects of various system parameters on the electroelastic response are explored. The analytical model developed in this work can be used to predict and optimize the multiphysical system dynamics with very good accuracy and substantially improved computational efficiency as compared to finite-element modeling using commercial packages.

Significant technological advances in sensing and communication promote the use of large sensor networks to monitor structural systems, identify damages, and quantify damage levels. Prognostics and health management (PHM) technique has been developed and applied for a variety of safety-critical engineering structures, given the critical needs of the structure health state awareness. The PHM performance highly relies on real-time sensory signals which convey the structural health relevant information. Designing an optimal structural sensor network (SN) with high detectability is thus of great importance to the PHM performance. This paper proposes a generic SN design framework using a detectability measure while accounting for uncertainties in material properties and geometric tolerances. Detectability is defined to quantify the performance of a given SN. Then, detectability analysis will be developed based on structural simulations and health state classification. Finally, the generic SN design framework can be formulated as a mixed integer nonlinear programming (MINLP) using the detectability measure and genetic algorithms (GAs) will be employed to solve the SN design optimization problem. A power transformer study will be used to demonstrate the feasibility of the proposed generic SN design methodology.

Based on our previous work on Mobile Actuator and Sensor Network, Applied Fractional Calculus, Sensor Networks and BUMMPEE (Bottom-Up Modeling of Mass Pedestrian flows implications for the Effective Egress of individuals with disabilities), a general framework is proposed for modeling and managing Mass Pedestrian Evacuations (MPE) in this paper. A distinctive feature compared with previous work is the incorporation of Individuals with Disabilities (IwDs) in understanding modeling and control of mass pedestrians evacuations. Networked Segway Supported Responders (NSSR) have been firstly employed in the research of modeling and control/managing problem of crowd pedestrians as mobile sensors and mobile actuators. Future simulation and experimental results will be referenced for public policy professionals and planners for better evacuation policy making and route planning.

The prevailing practice of design for mass customization manifests itself through a configure-to-order paradigm, which means to satisfy explicit customer needs and built upon legacy design. With pervasive connectivity and interactivity of the Internet and sensor networks, personalization has been witnessed in a number of industry sectors as a promising strategy that makes the market of one a reality. This positioning paper envisions an extension to design for mass customization and personalization (DFMCP). By exploiting implicit market demand information and revealing latent customer needs, DFMCP aspires to assist customers in making better informed decisions, and to the largest extent, to anticipate customer satisfaction and adapt to customer delight. Based on a multiverse of user experience, product differentiation and co-creation, the key issues of DFMCP are discussed.

An embedded sensor network is a network of sensor nodes deployed in the physical world that interacts with the environment. Each sensor node is a physically small and relatively inexpensive computer that has one or more sensors. These sensor nodes are often networked, allowing them to communicate and cooperate with each other to monitor the environment. Typically, an embedded sensor network is controlled by its own applications that can access the sensor nodes within the network. On the other hand, the sensor nodes cannot be easily accessed by applications outside of the network. Moreover, even within the same network, different applications might encounter a race condition when they are trying to access a sensor node simultaneously. The issue is related to system management. However, not much research has been done with a focus on the management of sensor nodes. In the past few years, Cloud computing has emerged as a new computing paradigm to provide reliable resources, software, and data on demand. As for resources, essentially, Cloud computing services provide users with virtual servers. Users can utilize virtual servers without concerning about their locations and specifications. With such an inspiration, this paper proposes a system, Sensor Agent Cloud, where users can access the sensor nodes without worrying about their locations and detailed specifications. Sensor Agent Cloud virtualizes a physical sensor node as a virtual “sensor agent”. Users can use and control sensor agents with standard functions. Each sensor agent operates on behalf of its user. The mandatory coordination of these sensor agents is related to the system management. Therefore, Sensor Agent Cloud must be an autonomic system that manages itself with minimum human interference. In addition, Sensor Agent Cloud supports international standard technologies regarding programming and agent communication (C and IEEE FIPA standard). Thus, it is expected that the proposed Sensor Agent Cloud can enhance the applicability and usability of embedded sensor networks in many application areas.

In this paper we present a traffic simulation system using traffic data gathered from readings by GPS devices installed on board of busses, i.e., busses running on the road are used as a mobile sensor network. Traffic data are inferred from the comparison of data about normal speed and current speed in a certain position. The simulation system was developed to invent mechanisms that use such information for traffic optimization. In particular, simulations allow to manage traffic lights in real time and to control the traffic system in order to give priority to public transportation systems.

Significant technological advances in sensing and communication promote the use of large sensor networks to monitor structural systems, identify damages, and quantify damage levels. Prognostics and health management (PHM) technique has been developed and applied for a variety of safety-critical engineering structures, given the critical needs of the structure health state awareness. The PHM performance highly relies on real-time sensory signals which convey the structural health relevant information. Designing an optimal structural sensor network (SN) with high detectability is thus of great importance to the PHM performance. This paper proposes a generic SN design framework using a detectability measure while accounting for uncertainties in material properties and geometric tolerances. Detectability is defined to quantify the performance of a given SN. Then, detectability analysis will be developed based on structural simulations and health state classification. Finally, the generic SN design framework can be formulated as a mixed integer nonlinear programming (MINLP) using the detectability measure and genetic algorithms (GAs) will be employed to solve the SN design optimization problem. A power transformer study will be used to demonstrate the feasibility of the proposed generic SN design methodology.

This paper presents a Web-based data inquiry and real-time control of sensor’s operating mode for structural health monitoring sensor networks. The main objective of the presented system is to provide a Web interface for real-time sensor data visualization, sensor-level damage diagnosis, and control of sensor’s operating mode. Web services are available both on distributed sensor nodes and a data repository machine. Users can request Web pages hosted on the sensor nodes or the data repository machine by specifying corresponding sensor IDs. The ability of directly accessing data on sensor nodes via internet allows users to monitor a structure’s performance in a timely manner. The damage diagnosis algorithms implemented on the sensor nodes help users to assess the structural health conditions without the need of transmitting sensor data to a central data station. The presented system also provides the capability of dynamically changing sensor’s operating mode through the Web interface. This feature greatly enhances the flexibility of the system to accommodate different sensing needs and achieve a long lifespan. The system has been tested in the Laboratory to validate its capabilities.

Considering the environmental load of material and energy consumption, many approaches to product design and manufacture have been discussed from an ecological viewpoint. In this paper, we propose the architecture of a sensor network system that promotes condition-based maintenance and reuse in order to reduce the environmental load in product manufacture and use. The system architecture stands on an agent-based sensor network and a modular reusable design. In this system, sensor data are used not only for condition-based maintenance but also for reuse management since they enable the estimation of product conditions that are indices for deciding a lifecycle option and a reuse plan. As a simple realization model of the proposed architecture, a motor-assisted bicycle CBM and reuse support system is developed by using wireless small sensor network and network agent systems. The system monitors product conditions and searches the next owner of each module for promoting reuse and avoiding product waste as much as possible. Case studies show some practical use and merits of this architecture.

Swarm robotics is an innovative approach to the control and coordination of multi-agent systems that use naturally inspired swarm intelligent methods to perform tasks. A swarm based approach can decrease the complexity and the cost of designing a cooperative multi-robot system. This paper proposes a general engineering approach to develop a robotic swarm that focuses on how to synthesize an emergent behavior and the associated inputs to this end. We validate our methodology by engineering a swarm to simultaneously rendezvous on a stationary light source. Furthermore, we also considered the case when the light source is slowly moving. The design is simulated in NetLogo, an agent-based modeling software, and implemented on the MASnet robot platform. This work demonstrates the basic knowledge and tools required to engineer a robotic swarm.

This paper presents a new platform with a team of lab-scale networked mobile robotic manipulators (SumoMote) which merges a mobile manipulator with wireless mobile sensor networks. Many existing platforms built for mobile manipulation are big and expensive. Our SumoMote is built small and inexpensive for applications where quantity is more important than size. The hardware and software of the SumoMote will be described. Then two application scenarios will be presented to illustrate SumoMote’s capability in mobile sensor networks and how the added manipulator can help.

This paper proposes a simultaneous localization technique of mobile robot and pedestrian in ubiquitous sensor network. For the robot localization, a dead-reckoning system is developed wherein odometer, magnetic compass, and heading angle rate sensor are used. The novelty of dead-reckoning system developed in this paper is that it does not use acceleration in motion dynamic equation. Since the dead-reckoning system does not use linear acceleration, the system is not affected by high frequency noise, which is usually contained in the accelerometer measurement. For the pedestrian tracking, ubiquitous sensor network such as IEEE 802.15.4 is used. In this paper, it is also assumed that the relative direction of the pedestrian from the mobile robot is measured on the robot platform. Extended Kalman filter is used to integrate the sensor measurements. Simulation results will be presented to demonstrate the superiority of the proposed simultaneous localization technique.

Recent proliferation of sensor networks in various application areas has promoted real-time behavioral monitoring of various physical systems. The opportunity to use sensor generated data dynamically for improving speed, accuracy, and general performance of predictive behavior modeling simulation is of paramount importance. The present paper identifies enabling modeling methods and computational strategies that are critical for achieving real-time or near real-time simulation response of very large and complex systems. It also discusses our choices of these technologies in the context of sample multidisciplinary computational mechanics applications and describes two examples to demonstrate the feasibility of integrating real-time data with real-time simulation.