G. Bhat, Y. Tuncel, S. An, U.Y. Ogras
Arizona State University,
Keywords: open source hardware, wearable electronics, health monitoring
Summary:The Annual World Report on Disability reveals that 15% of the world's population lives with a disability. Diagnosis, treatment and rehabilitation of this population currently depends on the behavior observed in a clinical environment. After the patient leaves the clinic, there is no standard approach to continuously monitor the patient and report potential problems. The quality of life of this population can be improved significantly with the help of wearable internet-of-things (IoT) devices that combine sensing, processing and wireless communication capabilities within a small form-factor. However, widespread adoption of wearable devices has been hindered by three major challenges. First, small form-factor IoT devices must operate under extreme energy constraints, since large and heavy batteries are prohibitive and flexible printed batteries have low capacities. Second, conventional rigid devices are uncomfortable to wear for long periods of time. Finally, the value of wearable IoT devices must be demonstrated by high-impact applications. To address these challenges, we propose physically flexible energy harvesting wearable IoT devices. Furthermore, we demonstrate our solution with human activity and gesture recognition applications that can benefit patients with movement disorders. Flexible hybrid electronics is an emerging technology that combines the performance advantages of rigid integrated circuits and form factor advantages of printed electronics. Using the FHE technology, we designed an open-source wearable IoT device that consists of a microcontroller, motion sensor, flexible photovoltaic (PV) cell, and energy harvesting circuitry mounted on a flexible substrate. As a result, it can be worn as a patch on the body. We envision that the wearable prototype and extensions to it can be used to create an open source ecosystem for health monitoring. Small form-factor wearable devices have to operate under tight energy budgets due to their small size. To address this, we consider long term recharge-free operation of wearable devices. Specifically, we harvest ambient energy to enable energy-neutral operation i.e. we aim to start and end each day at the same battery level. We achieve this by first modeling the energy harvested by flexible photovoltaic cells considering the effect of bending on PV cell efficiency. Then, we develop a dynamic programming algorithm that optimally allocates the harvested energy while maximizing the utility to the user. These algorithms enable energy-neutral operation by maximizing the energy harvested by flexible PV cells and optimally allocating the harvested energy. Finally, we demonstrate our wearable device solution with human activity and gesture recognition applications. Human activity recognition is important for personalized health care as it allows the doctors to understand patients' daily activity patterns before prescribing therapy. Similarly, gesture recognition can be used for gesture-based control and interaction with assistive devices. We implement these applications on our wearable device and conduct user studies to demonstrate their applicability. Experiments with user studies show that the human activity and gesture recognition applications achieve 97% and 98% accuracy, respectively. If our paper is accepted, we will perform live demonstrations of human activity and gesture recognition applications with our energy harvesting wearable device.