Search

Researchers have extensively explored indoor localization in recent years. Robotic applications often require precise positioning, which is difficult when access to a GPS is limited or compromised. Many of the existing approaches require either prior knowledge of the local environment and fixed landmarks or complex and expensive hardware to achieve the necessary degree of accuracy. In this article, we examine the use of the HTC virtual reality hardware along with our novel approach, VIVEPOSE, to perform indoor robotic localization. We then compare the accuracy of our proposed indoor localization system to current approaches that use traditional odometry sensor-based localization. Finally, we present and demonstrate a leader-follower approach using VIVEPOSE and show how the HTC Vive tracker can be successful for indoor localization for a robotics application.

Background Gut-directed hypnosis (GDH) is a complimentary therapy for Inflammatory Bowel Disease (IBD), that can be learnt by patients to practice self-hypnosis. GDH in IBD has augmented remission and improved inflammation. GDH has a history of successful use for Irritable Bowel Syndrome (IBS). In IBD it may also improve IBS-like symptoms in remission and recovery from surgery. GDH is suitable for youth and adult IBD patients. In hypnosis, a relaxed state is inducted then suggestions to subconscious mind processes are made. In IBD, the mechanism of action of GDH is unknown but may influence the disease stress response. Aims Aims are the development of a GDH self-hypnosis protocol for IBD, with appropriate target symptoms. Patients first learn to practice with a clinician, then as complimentary psychotherapy for remission augmentation, IBS-like symptoms, and surgery recovery. Methods GDH is practiced first with a clinician, and then by patients as self-hypnosis (table 1). Patients receive psycho-education on GDH for IBD. Next, appropriate treatment goals are made, based on target symptoms. Relaxation techniques induce patient to a deeply relaxed state. Therapeutic suggestions specific to patient goals are given: verbal suggestions, visualizations, and post-hypnotic suggestions. Suggestions can focus on having a healthy digestive system, inflammation and symptoms reduction, and achievement and sustainment of remission. Patients emerge from hypnosis, are debriefed, and encouraged to practice ongoing self-hypnosis. Results In IBD, GDH self-hypnosis can be learnt from clinicians and practiced by patients as a complimentary therapy. Patients’ achievement and sustainment of remission, with clinical markers of inflammation can be monitored. Patients can monitor subjective improvement of IBS-like symptoms and post surgery, recovery progress can be monitored. Conclusions GDH has a history of use for IBS. In IBD, it has been shown to modulate remission, and may improve IBS-like symptoms, and in surgery recovery. The mechanism of action of GDH in IBD may influence the disease stress response. Clinicians trained in GDH are limited currently. Patients may learn GDH self-hypnosis as a complimentary psychotherapy. Funding Agencies None

To meet new needs and to respond to changes in the energy market, Hydro-Quebec Trans Energie (HQT) under-´ takes an important research project, named PRIAD, to improve existing tools for asset management and modelling system. The aim of this project is to assess the impact of the unavailability of equipment, such as breakers, sectionalizers, physical buses, power transformers, etc., on the performance of the Hydro-Quebec (HQ) network. Power system node-breaker (PSNB) models are of common use in operation, while bus-branch models are usually used in planning studies. PRIAD relies on a transmission system reliability simulator named PRISME. PRISME is HQ’s first planning application that requires a PSNB model. To this end, we have developed an algorithm to convert the traditional bus-branch network model to a detailed node-breaker representation. To generate the node-breaker model (PSS/E format), we used the state estimator file in IEEE format, the Energy Management System (EMS) network connectivity model (CIM format), and the state of switches from the Supervisory Control and Data Acquisition System (SCADA) file.

Formation control has been a popular research topic in recent years. This is primarily due to the the use of formation control to coordinate the movement of a swarm of robots. In this paper, with the primary objective of guaranteeing image overlap, a novel bird flocking inspired formation control algorithm was developed to control a swarm of unmanned aerial vehicles (UAVs). Our approach uses only stereo cameras, global positioning system (GPS), and inertial measurement units (IMUs), and removes the need of feature or pattern matching. By utilizing stereo cameras, virtual tracking points can be calculated. The convergence of these tracking points are chosen as the primary formation control goal because if the virtual tracking points are converged, it is guaranteed that the field of view of the cameras of the various UAVs always have some overlap. This allows for different types of stitching methods to be used to form a larger image with no gaps, depending on the configuration of tracking points. Five different control algorithms were developed: direct method, proportional integral derivative, bang-bang, short window model predictive, and long window model predictive. The formation controller was tested in three simulation environments to verify the robustness and the effectiveness of our algorithm.

Cloud computing is now a global standard computing topology and has been widely studied for many years. Less frequently researched is the use of cloud and edge computing to optimize the performance of a system as a whole. One important aspect of cloud and edge computing is managing the placement of the applications in the network system so as to minimize each application’s runtime, given the resources of system’s devices and the capabilities of the system’s network. The properties of containerized applications now make this possible. The process of containerization creates a lightweight, mobile, packaged application for each of the algorithms in a system. These applications can then be deployed easily and quickly on any layer of the cloud and edge computing architectures. In this research, a fuzzy placement control system is designed to place applications on the cloud-edge model. As verified by simulation, the fuzzy placement control system proposed reduces the total runtime of the applications by placing applications in an efficient location.

Hydro-Quebec TransEnergie (HQT) has been employing predictive modelling methods to manage its assets for over a decade. To meet new needs and to respond to changes in the energy market, HQT undertakes an important research project in order to improve existing tools for asset management and modelling system. In this paper, we present a quick review of the global asset management model at HQT. Furthermore, we introduce a Contingency Analysis (CA) approach which will be integrated in the reliability simulator module of the HQT global asset management model. To this end, we augment the traditional bus-branch data model to provide a detailed node-breaker representation that contains detailed nodes, breakers and other switching devices. Then, we analyze the impact of the equipment unavailability on the network behaviour using a detailed CA approach. This contingency analysis approach not only informs the asset management engineers in case of violations, but also suggests several remedial actions to eliminate the violation.

Computer Vision is a valuable tool that can be used to enhance the components of many working systems. In particular, many applications can be improved by incorporating instance segmentation into their designs to help better process visual information in the surrounding environment. However, instance segmentation algorithms have traditionally been too slow to be used by any real-time systems that could benefit from using them. This includes examples such as self-driving vehicles or autonomous drones. In this work we provide an overview of the shortcomings for current instance segmentation algorithms, introduce an ongoing effort to create a new one that achieves ultrafast speeds without sacrificing competitive accuracy, and explain the advantages of employing an ultra-fast real-time version of one as a component in different systems.

The appeal of low cost computing and on demand scaling of cloud technologies has resulted in the migration of many software applications to the cloud. This increased reliance on the cloud translates to a direct dependence on the cloud data centers, which form the modern cloud. These data centers are complex buildings composed of many system of systems that interact to host the end applications. Detecting anomalous events in this system of systems and then identifying their root cause in a timely manner is a demanding task. In this paper, we present a graphical model to encapsulate the cloud data center system of systems and share a method for reducing the search space for root cause analysis.

Background IBD patients have a heightened risk for mental health illness, but general psychotherapy has shown mixed results. This psychotherapy protocol specialized for IBD patients uses recommended mental health therapies to treat specific chronic mental health illness. Therapy focuses on practicing CBT, clinical hypnosis, and emotion regulation healthy coping skills to self-manage chronic mental health symptoms. Standard therapies for acute mental illness, e.g. anxiety, depression, and suicidality, should be used as appropriate. Aims Aims of psychotherapy in IBD include improvement of mental health, symptoms management, quality of life (QOL), and adherence to medical treatments. Cognitive behavioural therapy (CBT) changes maladaptive coping behaviours and thinking. Clinical hypnosis improves somatic symptoms, e.g. chronic pain, nausea, and cramping. Emotion regulation improves ability to process stressful emotions, and has been shown to be affected by chronic diseases, including moderate to severe Crohn’s disease. Methods Patients learn to self-manage chronic mental health symptoms with healthy coping skills. CBT allows patients to identify maladaptive coping behaviours and thinking. Healthy coping behaviours and thinking patterns are chosen, supported, and adapted to. Clinical hypnosis uses relaxation techniques, e.g. progressive muscle relaxation, visualizations, and positive suggestions as a complimentary therapy for somatic symptoms. Emotion regulation develops tolerance and healthy processing of stressors. Results Therapy effectiveness is evaluated by improvements in patients’ mental health, symptoms management, QOL, and adherence to medical treatments. Mental health improvement, QOL, and emotion regulation are monitored by patient self-report, e.g. questionnaires. Maladaptive coping behaviours and thinking changed using CBT are individual to each patient. Somatic symptoms improved using clinical hypnosis are determined by monitoring or patient self-report. Adherence to medical treatments, as appropriate, is monitored. Conclusions This IBD psychotherapy protocol uses gut-directed CBT, clinical hypnosis, and emotion regulation therapies, designed to improve mental health, symptoms management, QOL, and adherence to medical treatments. Healthy coping skills treat maladaptive coping behaviours and thinking, somatic symptoms, and emotion dysregulation. Chronic mental health illness associated with IBD, is treated using this specialized psychotherapy. Funding Agencies None

Cloud computing infrastructures have become the de-facto platform for data driven machine learning applications. However, these centralized models of computing are unqualified for dispersed high-volume real-time edge data intensive applications such as real time object detection, where video streams may be captured at multiple geographical locations. While many recent advancements in object detection have been made using Convolutional Neural Networks, these performance improvements only focus on a single, contiguous object detection model. In this paper, we propose a distributed Edge-Cloud R-CNN pipeline. By splitting the object detection pipeline into components and dynamically distributing these components in the cloud, we can achieve optimal performance to enable real time object detection. As a proof of concept, we evaluate the performance of the proposed system on a distributed computing platform including cloud servers and edge-embedded devices for real-time object detection on live video streams.

The use of Blockchain in the Internet of Things (IoT) promises to provide an avenue of decentralized, fault-resistant management and data immutability. The wide range of research on blockchain, focused on how to improve upon its drawbacks of latency and resource consumption, demonstrates that the blockchain design has flexibility. Almost every aspect of blockchain can be tailored to fit the requirements of a desired application. However, the wide range of options in configuring blockchains presents challenges in the adoption of blockchains to secure IoT. In this paper, we present the opportunities and challenges of implementing blockchains and present a use case of integrating blockchain into an IoT framework for securing sensor data acquisition.

Deep neural networks (DNNs) are currently among the most commonly used machine learning methods in content understanding such as computer vision and natural language understanding (NLU). One of the best characteristics of these methods is their modular design – the ability to change the connectivity patterns of layers, try different activation functions, inject different statistical approaches such as normalization and dropout in the network, and many other actions – in every aspect of deep learning networks. While the majority of deep learning applications simply use cross-entropy, L 1 , and L 2 losses, subjective loss function can actually result in impressive performance improvement. In addition, architecting the last layer of DNNs – referred to as the prediction layer – according to the needs of the application increases the discriminative power of the DNNs. This paper aims to investigate how particular choices of loss functions and prediction layer architecture affect deep neural networks and their learning dynamics, as well as the robustness of various effects. Furthermore a real-life application to measure customer loyalty called Deep Net Promoter Score (DeepNPS) from online product reviews is also proposed. The results are promising for learning more latent features and matching the customer feedback with the NPS score.

Virtual reality (VR) has become a very popular gaming platform in recent years, but has not been used as much in the research domain. Several technologies used to create virtual environments can also be used to assist in research. For example, recent developments in VR allow for easier integration of humans in the loop for control and manipulation of complex, multi-modal system of systems. In this paper we show some of the methods in which a virtually emulated system can be integrated into an artificial intelligence (AI) based system of systems. We demonstrate how this integration allows the reseacher to more easily understand and analyse the inter-system dynamics, and to improve performance using post-simulation visualization, real-time simulation interaction, hardware emulation, and physical object tracking. The results of using virtual reality as a simulation environment shows the usefulness of the tool to system of systems researchers.

Modern applications of Unmanned Aerial Vehicles are increasingly attracting the attention of traditional safety and reliability fields. There exist many standard approaches for determining UAV fault detection. However, there doesn't exist a method that is not only model independent but also has the ability to detect faults which have not been predefined for the UAV system. In this research we present two supervised machine learning algorithms implementing Logistic Regression and Linear Discriminant Analysis of Algorithms, respectively, to predict UAV faults. The data which has been used for these approaches comes from discrete-sampled, de-noised analog signals based on the voltage and current inputs belonging to four actuators of the UAV drones. In addition, we demonstrate that by using a five-fold cross validation process to generate different types of training and test datasets, the optimized model can be selected. We verify our results through an analysis describing the accuracies of our proposed model.

Background: Glioblastoma Multiforme (GBM) has well documented systemic and local immunosuppressive mechanisms to escape immune surveillance and grow. GBM tumor cells as well as the microglia within it have a high incidence of PD-L1 surface expression which makes it more susceptible to anti-PD-L1 antagonism and ADCC through avelumab therapy. Methods: This is a single center, phase 2, open label, add-on, single dose study of 156 weeks duration in patients receiving standard therapy for newly diagnosed GBM. In total 30 patients will be entered into the study within 3 weeks of finishing their last day of combined radiotherapy/temozolomide. The following are the results of the first interim analysis completed when the first eight patients completed 52 weeks or an end of study visit. Results: 24 patients have so far started therapy. There as been no unexpected treatment emergent adverse event (TEAE). Two patients transiently withheld therapy because of immune related TEAE’s and none permanently. The objective response rate at week 52 for the first eight patients was 50% with 2 (25%) having a complete response and 1 (12.5%) a partial response. Conclusions: These preliminary results suggest that the addition of avelumab to standard therapy in patients with GBM is safe. Efficacy trends look promising.

Modern mobile devices are resource limited. We can utilize cloud computing to ensure optimum utilization of mobile device resources. Energy consumption and device latency can be reduced significantly by computation offloading. To ensure optimum utilization of mobile device resources and environment resources such as bandwidth and latency, computation offloading must be done in a strategic way. We have extended our work on a previous paper by proposing a dynamic algorithm based on machine learning which considers network topologies and device resources and makes a decision to offload computation to the cloud. The algorithm adopts with changes in environment and device parameters at run-time. We have designed a framework that resides both in the device, and in the cloud. The device framework monitors device and network parameters and based on user activity decides whether or not to offload computation to the cloud. When the device and network parameters crosses a certain threshold then it sends data to the cloud server using python sockets. The cloud framework receives the data and runs the app in the cloud. After executing the app in the cloud, the framework retrieves the output and sends it back to the local device. The device framework then gives the data to the app and the app completes its execution. We have performed a simulation of the system in a cloud server, and showed that energy can be saved by computation offloading. We also present a financial estimation to calculate the cost of computation offloading.

Pedestrian recognition is a key problem for a number of application domains namely autonomous driving, search and rescue, surveillance and robotics. Real-time pedestrian recognition entails determining if a pedestrian is in an image frame. State-of-art pedestrian detection convolution neural networks(CNN) such as Fast R-CNN depend on computationally expensive region detection algorithms to hypothesize pedestrian locations. This paper presents a simple, fast and very accurate approach by cascading fast regional detection and deep convolution networks. Convolution networks have been shown to excel at image classification. However, convolution networks are notoriously slow at inference time. In this work, we introduce a fast regional detection cascaded with deep convolution networks that enables real-time pedestrian detection that could be used to alert a driver if a pedestrian is on the roadway. The classification CNN has given an accuracy of 95.7%, with a processing rate of about 15 frames per second on a low performance system without a graphical processing unit (GPU).