Diagnostic and Monitoring Tool for Amputee Patients Wearing a Prosthesis

In this project, we propose a universal internet of things (IoT) device that connects between the residual limb of an amputee patient and the prosthesis to measure the patient’s comfort levels. This device's novelty lies in its ability to physically connect to most of the currently available prosthetic devices to provide the user with insightful and relevant data such as the pressure points, the wear time per day, or feedback on movement efficiency.

A Metrics-Based Approach to Support Sustainable Product Development and Consumption

Sustainable product development adheres to the needs of the present while developing a product without compromising the ability of future generations to meet their needs. This proposal aims to develop a suite of sustainability metrics that can be computed by analyzing data like company website, product material information, etc. The metrics will be integrated into the existing Arbor sustainability product suite via dedicated mobile app and browser extension.

Improving Donor Relations through Machine Learning

This project involves looking at machine learning as a new way to engage with donors for charities that can result in a more effective channel of communication.

Low Cost Eye-Tracking Data Analytic System for Vision Therapy

A significant percentage of children around the world are affected by vision problems, such as lazy eye, that cannot be fixed by eyeglasses. When left untreated, vision problems may lead to learning difficulties. Vision therapy offers treatment to many of these problems where eye-trackers are tools used to diagnose and measure the progress of patients undergoing vision therapy for these vision problems. However, vision therapy remains out of reach for many patients due to the cost of eye-tracking services.

A Cross-modal Video Representation Learning Framework for Query Retrieval and Automatic Trailer Generation

Grokvideo Inc. works on developing new technologies for extracting the best possible information from video content. This research project will provide new solutions to text-based query-video retrieval and automatic trailer generation for eventful videos such as movies and serials. Current methods seem to work well for short, non-eventful videos, say documentaries, but fail otherwise.

Development of Models and Control Schemes for Soft Open Points in Distribution Systems

More than 80% of power outages in power grids are caused by faults in distribution networks. To improve the system’s reliability and resiliency, the service restoration is critical after faults. The Soft Open Point (SOP) is an emerging power electronic device, which can be connected to terminals of feeders or between networked microgrids, and can be used to realize service restoration. In this proposal, we aim to develop models and control schemes of SOPs in distribution systems for service restoration purpose.

Secure Communication Based on Public Key Distribution with Imbalanced Twin-Field-QKD-like Infrastructure in Classical Coherent Optical Fiber Network

In quantum key distribution (QKD), the twin-field has been proposed as a means to overcome the linear key ratebound, and hence the key generation rate, as well as to enable repeaterless QKD transmission. Despitesignificant progress in TF-QKD and its variations in recent times, the key rates are still significantly low comparedto practical optical fiber networks.

Stored-Grain Bin Monitoring by Electromagnetic Imaging

The safe storage of grains is crucial for the food supply worldwide; for example, the storage loss is estimated to be between 2% to 30% depending on different geographic locations. In this project, an advanced signal processing algorithm (a deep learning approach) is developed to enhance the identification process of the moisture contents (MC) of grain bins from the measured electromagnetic data. This deep learning approach for grain bin monitoring will significantly accelerate the identification process of the MC as compared to existing techniques.

Using AI to Help First Responders Assess Skin Burns

Burns are a common type of skin injury that cause numerous deaths around the world every year. Timely assessment of burns plays an important role in a successful treatment. Traditionally, burns are assessed through visual and tactile observation by clinicians. This method of assessment is highly inconsistent as it depends on the availability of a clinician and the clinician’s level of experience. Deep convolutional neural networks (CNNs) have the potential to offer an alternative for burn evaluation that is accurate, fast, inexpensive, and can be performed easily by the first-responders.

Exciton-Induced Aggregation in Novel Guest Emitters for wet-coated OLEDs

The promise of low-cost fabrication via “wet” processes, such as inkjet printing, in manufacturing flat panel displays and solid-state lighting has long been one of the main motivations behind the pursuit of Organic Light-Emitting Devices (OLEDs). The vast majority of OLEDs in commercial products are still fabricated by costly “dry” vacuum-deposition methods however. A primary reason is the significantly lower electroluminescence (EL) stability of devices made by wet-coating in comparison to their vacuum-deposited counterparts.