.

Sunday, July 21, 2019

Wearable Motion Detection Technology to Detect Falls

Wearable Motion Detection Technology to Detect Falls Louise Patterson 1. Project Title The development of a fully working model using wearable motion detection technology to detect and alert to falls within the home. 2. Abstract The focus of this document concerns the research and methodology into the problem of people with care needs being able to live independently within their own homes with minimum intervention. Elderly people particularly tend to have an increased risk of falling and sustaining serious injury than younger people. Other people with certain medical conditions also have a higher risk of falling due to seizures, balance problems etc. Current fall detection alert systems are expensive and obtrusive. They require landline phone connections and a call centre network with access to personal information to contact the appropriate carer. Sensors are rarely waterproof and can be easily damaged. The way the sensors are carried on the body can irritate and become annoying to the patient. One of the biggest, most costly failures of current systems occur when a call button has been touched by accident, resulting in accidental alerts being sent to the call centre. The main objective of this research and proposed model is to provide an investigation into the possibility of using new technology to provide a more effective system. Which Such a system will work only for fall detection and send alerts to appropriate personnel with the option for camera access to enabling them to view situation on way to alert. Specifically the project seeks to discover if new ring motion detection technology can be used to provide a more cost effective way of detecting falls within the home. Provisioning of a sensor which is both waterproof and unobtrusive to the wearer, which can be used to send an alert via txt or email directly to carers, thereby removing the need for an expensive, dedicated landline alert system and call centre network. Table of Contents (Jump to) 1. Project Title 2. Abstract 3. Aims 4. Objective 5. Justification 6. Literature Review 7. Methodology 8. Work Plan 9. References 3. Aims The aim of the project is to develop a working model for fall detection using motion detection technology in the form of a wearable ring, thereby replacing current, outdated and non user friendly equipment. Thereafter presenting the working model to appropriate bodies in order to facilitate a huge reduction in what has become a costly and awkward system to run and maintain for all parties. To achieve this, research in what defines a fall will be carried out as well as research for application creation using the processing language and motion detection programming using the available software development kits provided by the ring developers. Current fall detection applications will be researched to compare and improve on proposed model. To achieve this, research will be carried out to determine what defines a fall in the current system, as well as the range of fall detection applications currently available. Following this, research will be required into the most appropriate type of application creation tools to use in the working model. Tools include the type of processing language required and the particular motion detection programming available through the software development kits provided by the ring developers. Comparisons will be drawn from research findings in order to discover what improvements could be made on the proposed working model. 4. Objectives 5. Justification 6. Literature Review The purpose of the papers read is how to define the fall within the scope of the project and what technologies are best suited to project. With this in mind the following papers contain information on fall detection and the different technologies that may be used to complete a project of this type. Challenges, issues and trends in fall detection systems by Raul Igual, Carlos Medrano, and Inmaculada Plaza: Published online July 6 2013 in the BioMedical Engineering OnLine Fall detection system using Kinect’s infrared sensor by Georgios Mastorakis and Dimitrios Makris : Published December 2014 in the Journal of Real-Time Image Processing Heterogeneous multi-sensor fusion based on an evidential network for fall detection by Paulo Armando CAVALCANTE AGUILAR, Jerome Boudy, Dan Istrate, et. Al.: Submitted to HAL on the 14 Dec 2012. Accurate, Fast Fall Detection Using Gyroscopes and Accelerometer-Derived Posture Information By Qiang Li, John A. Stankovic, et. Al. Challenges, issues and trends in fall detection systems: this paper covers many of the different studies already taken place. It details the devices and methods used to detect falls. Fall detection system using Kinect’s infrared sensor : This paper cover the use of the xbox’s kinect sensors in fall detection. The same technology used in the nod ring. Heterogeneous multi-sensor fusion based on an evidential network for fall detection : covers how the network is used to alert and detect the falls. Accurate, Fast Fall Detection Using Gyroscopes and Accelerometer-Derived Posture Information: covers how the bodies position is used with accelerometers and gyroscopes to detect falls. 7. Methodology The research required to enable completion of the project will the follow the flow diagram used within Objectives. Secondary research will be carried out into what defines a fall. This will require papers, articles, books and websites to be read on the definition of what constitutes a fall. The information gained from this research will be compared and contrasted in to tables to analyse what movements and positions best constitute a fall and will cover the most appropriate scenarios for the system model. During this research any currently available algorithms, diagrams and coding will be collated, analysed and used to determine the fall definition selected for the project. Another aspect required for the system model is the final position of the fall and the time of inactivity that would indicate a fall. The data will be collated and analysed using graphs and diagrams to select the most appropriate solution that encompasses the most data to create an all-encompassing fall detection system. Time inactivity data will be collected from research data to compare and contrast time retrieved to select the median figure for inactivity and use this in the development of the system. Then all the collated data will be compared analysed and a conclusion drawn, showing the best options for the project, and the development of the equations required for coding of the device sensors. Once a fall definition has been selected the language used to program the device sensors needs to be researched and learned. The sensor device has available software developer’s kits (SDK) for android, iOS and UNIX among others. The use of these kits will require a steep learning curve and require reading and taking practical courses on learning the languages required. Learning the language, will be achieved through books, online courses and available API’s and code. The Project Supervisor will assist in the project ensuring the correct language is used and coding is fit for purpose. Although the ring comes with its own software, other methods of programming appropriate to the current skill set will be researched, by contacting developers through the developer’s forum and other programmers to assist with programming the sensors. At this stage any compatibility issues with the sensor and devices will be noted and used during the testing of the completed model to facilitate improvements. During the process of learning the coding language, the programming of the ring using this knowledge will be initiated. Sample codes will be modified and tested by programming the device to test functionality. Current fall technology research will include investigating current papers, applications and systems. Available current applications will be compared against the proposed project model to determine possible areas for improvement. Initial alert sending programming will be programmed within the rings SDK, producing results of positive fall detection and sending them to a receiving device. Knowledge and programming developed during research will be used to complete alert programming. Testing will be carried out using simulated alerts on various platforms, at different distances from sensor to base station. Results will be collated and displayed in graphs showing comparisons and analysis. The alert sending system which sends messages to mobile phones and emails will be programmed using the SDK or language used to program other functions. The completed model is to be tested using actual falls which creates an ethical issue. To address this issue local martial arts instructors will be approached to request their assistance with testing. The instructors will be asked to provide the test equipment to classes in order to record falls that take place during normal class activities. To fulfil any ethical considerations these tests will be recorded and the appropriate paperwork for consent obtained. These tests will ensure fall detection, alerts sending and receiving and the effective range of the system. The results will then be collated and reproduced in graphs, tables and illustrations to demonstrate success and failure rates, distance achieved and sending and receiving times. Future developments are investigated through searching for additional functionality of the sensor devices used within the model and possible cost effective additions which, when added to the system model improve the project and add features resulting in a reliable care model. The final step will be compiling the report for submission. This will done by pooling all research, tables, graphs and coding into the documentation required for the MSc project. 8. Work Plan The MSc Smart Networks course leader has stated the project must be completed and submitted for marking so graduation can be achieved in December or January. With this is mind the completed project must be submitted to lecturer in plenty of time to allow for marking and external evaluation. To ensure this is the case a final submission date of October 30th 2015 has been selected with a start date of the 22nd of June 2015. A draft submission is to be submitted no later than the 25th of September. Timeline As a literature review is a learning and development process this will be an ongoing method from June to the end of September. Programming and initial testing will commence in the first week in July and continue until a completed model is ready for final testing. Final testing is to commence in the third week of September and last for two weeks. Compiling of data and research will be compiled as an ongoing process, however writing the project will start in earnest at the same time as final testing. Due to the complexity of the project fortnightly meetings / tutorial sessions will be arranged with the supervisor. 9. References

No comments:

Post a Comment