Port, Stuart (2022) Development of object detection system to minimise tripping and slipping risk. Research Master thesis, Victoria University.

Abstract

Falls within the elderly community are estimated to have cost the Australian Healthcare system approximately $498.2 million in 2001 and are predicted to rise threefold to $1375 million in 2051. A vast majority of these falls are a result of tripping, slipping, and stumbling hazards. These types of falls are commonly caused by everyday household objects such as extension cords, shoes, children’s toys, and rugs. The primary reason for these falls among the elderly is the decline in vision and gait functionality. The wearable technology industry has seen unprecedented growth in recent years, with a market value of $28 billion USD in 2020, and costs are expected to reach $74 billion USD in the year 2026. The current wearable market mainly consists of wearable smart watches and smart glasses, with a vast majority being used in fitness and healthcare applications. A small majority of wearable devices have been introduced into the age care industry, with smartwatches being used to detect falls and alert carers when an individual has had a fall. These devices have only recently been introduced to the market as they rely on low-power cellular components and, as a result, lack the ability to predict and prevent falls. With advances in the Artificial Intelligence industry and the introduction of the tensor processing unit, there now exists practical solutions to implement Artificial Intelligence onto mobile and internet of things devices. This has allowed real-time object detection to be implemented in real-life applications. An area in which these improvements can be taken advantage of is assisted walking for the vision impaired and robotics and assisted walking devices for the elderly. Due to the lack of fall prevention devices, this research aims to investigate the feasibility of detecting hazardous objects and performing fall prevention from the waist by using a Convolutional Neural Network that is implemented onto an embedded system. In addition, the Convolutional Neural Network will be trained to detect hazards such as spills on various types of flooring. The results of this study will benefit the impaired vision community, robotics community and the elderly population, as it can lead to the development of assisted walking devices.

Search Google Scholar

Repository staff login