 |  |  |
| | | |
Intelligent Wearable Sensing Technology for Next-Generation Multi-Sports Training
We have developed an IoT (Internet of Things) framework for next-generation sports training. To validate its performance, a wireless wearable sensing device (WSD) based on MEMS (microelectromechanical systems) motion sensors was used to recognize different badminton strokes and volleyball spikes and classify skill levels from different players of these respective sports. The system includes a customized sensor node for data collection, a mobile app and a cloud-based data processing unit. The WSD developed is low-cost, easy-to-use and computationally efficient compared to video-based methods for analyzing different arm motions. It offers the advantage of dynamic monitoring of multiple players in any environment. As an example of application, experimental results show our system is capable of recognizing three different badminton actions, i.e., smashes, clears and drops, with an accuracy rate of 97%. in addition, the skill assessment function can differentiate between professional, sub-elite, and amateur players from their stroke performance. Our work aims to change the way of sports training from experience-driven to data-driven, and which can be easily extended to analyze the motions and skill levels of players in other sports (e.g., soccer, basketball, tennis, table tennis, squash, etc.) for training and/or practice.
Motion, Location, Indoor/Outdoor Positioning Solutions
CloudNav has created a suite of sensor fusion technologies that fuse inertial sensors, bluetooth beacons, and GPS to create wearable or motion products that can accurately understand human and non-human movement. Our indoor/outdoor positioning solution today is accurate down to 1 meter with future development to reduce that to less than .5 meters which is industry leading. CloudNav has enabled wearables for sports, children's activities, AR/VR, and healthcare with the ability to identify activities such walking/running/biking/sedentary/sitting/standing/laying along with body parts relative to the body. We currently have an iOS and Android based indoor location solution being deployed for retail customers and are working on a location wearable for indoor/outdoor applications where the user needs to be accurately monitored for instance elderly and blind individuals. We are also working on a IMU to accurately determine precise drone movement using off the shelf consumer sensors that automatically calibrates itself.
Deep learning in spacetime
physicsAI provides real-time machine perception in both space and time to enable autonomous systems to make intelligent decisions and adapt to new environments using end-to-end deep learning.
MRI MATH
We are developing a clinic-ready user-friendly software for radiologists and radiation therapists to view 3-dimensional segmentation of MRI brain tumor. The software is semi-automated, a desired feature in the clinic. The user interacts with an intuitive interface to check the labelling and modify it quickly, if needed. The quantitative measures are readily computed once the physician approves the final results. Our tool is multi-modal as it uses T1, T1 with contrast, FLAIR, and T2 MRI images. Clinical studies conducted on 11 patients over a period of 10 years show that our software is able to detect early recurrence of the brain tumors 3 years earlier than the practicing clinician. Our core algorithm won the runner-up Best Paper Award at the IEEE International Conference on Bioinformatics and Biomedicine in 2015 and was also a winner of the 2016 Multimodal Brain Tumor Image Segmentation challenge. The global MRI systems market is projected to reach $7.19 Billion by 2021 from $5.61 Billion in 2016, at a compound annual growth rate of 5.1%. In 2016, North America accounted for the largest share of the MRI systems market. A rising geriatric population is driving the demand for MRI systems, including software applications.
VerticalChange
VerticalChange is HIPAA-compliant and uses state-of-the-art Amazon Web Services (AWS) products where customer data is hosted in an isolated Amazon EC2 region using two Availability Zones (AZ). VerticalChange uses AWS monitoring services, including CloudWatch, to monitor the state of the infrastructure and alert us about security incidents. In addition to our own monitoring and audit of access logs, Clients can use event tracking within VerticalChange to review user activity. User activity is available on an individual user basis or for a single contact. Easy access to this information helps VerticalChange subscriber administrators manage staff and track the occurrence of data changes for individual contacts or records. VerticalChange also helps solve the complexity of data tracking and sharing in collaborative care partnerships or collective impact models with multiple partners from different industries. Eliminating data silos resulting common to fragmented programs, the software offers real-time access for all collaborators and the ability to evaluate impact and outcomes. Regardless of the data type (health, education, program, research, etc.), the system hosts large and varied datasets without compromising speed and efficiency. Finally, it can be easily integrated with multiple third-party applications, including data visualization software, to offer a comprehensive solution.
Nature inspired image processing software
Our technology is based on the innovative analogy that pixels in an image can be seen as an electric charge producing an electromagnetic potential and interacting electrically with the surrounding “charged” pixels. In electromagnetic theory, the behavior of electric fields along boundaries has been a well-known phenomenon for more than a century. The technology leverages this knowledge to infer different segmentation criteria used to recognized features (concavities, convexities, corners…) in the image and draw contours. An interesting point is that since electromagnetic laws can be transposed in any number of dimensions, the technology is compatible with 2D images, 3D depth maps and 3D point clouds. The capabilities of our technology extend beyond segmentation. Indeed, the electromagnetic analogy can also be used to identify, once an object is segmented, the most stable grasping points (including handles) if the object needs to be handled by a robot, an added advantage to penetrate the robotics market.
Portable Hardware Energy Optimisation (PHEO)
Software that collect energy use statistics for third party applications. PHEO then uses these statistics to drive its online optimisation process to produce code variants that both pass acceptance tests and use less energy. These variants are then presented to the developers for validation and release. Existing solutions imposed by operating systems are not able to improve the efficiency of apps, but just exert course-grained control, such as rationing of functionality. App developers are not skilled in energy-aware coding and typically do not have test platforms at their disposal. We aim to optimise energy consumption of third-party apps on smartphones. Our framework will identify issues and suggest fixes that reduce energy consumption in apps without affecting their functional requirements. Our framework will continually operate over app source code, tracking current performance on a range of phone hardware, system software, user configurations and use patterns.
Reconfigurable Physically Unclonable Function (PUF) Based Security
The demand for localized hardware security is continuously growing due to the rapid expansion of online distribution of interconnected networks and devices carrying critical/sensitive and personal information on shared networks. This technology is also the best known approach to prevent offline attacks of critical data, such as keys or complete memories. It uses a feature of crossbar architecture with integrated resistive random access memory (ReRAM or memristors) as a Physical Unclonable Function (PUF) and improves upon it using analogue tuning of the memristors’ conductances to maximize the PUF’s functional performance. The memristor based PUFs have a simple and relatively low-cost fabrication process, small footprint and compatible with complementary metal-oxide-semiconductor (CMOS) circuits. The instance reconfigurability make this technology superior to other PUF hardware and highly suitable for security applications. These hardware solutions have wide-ranging applications including securing payments, protecting highly sensitive data, for anti-counterfeiting and anti-cloning, to prevent identity theft, to prevent the piracy of media content and software apps and preventing software reverse engineering.
