Energy-Positive Activity Recognition - From Kinetic Energy Harvesting to Smart Self-Sustainable Wearable Devices

Abstract

Wearable, intelligent, and unobtrusive sensor nodes that monitor the human body and the surrounding environment have the potential to create valuable data for preventive human-centric ubiquitous healthcare. To attain this vision of unobtrusiveness, the smart devices have to gather and analyze data over long periods of time without the need for battery recharging or replacement. This article presents a software-configurable kinetic energy harvesting and power management circuit that enables selfsustainable wearable devices. By exploiting the kinetic transducer as an energy source and an activity sensor simultaneously, the proposed circuit provides highly efficient context-aware control features. Its mixed-signal nano-power context awareness allows reaching energy neutrality even in energy-drought periods, thus significantly relaxing the energy storage requirements. Furthermore, the asynchronous sensing approach also doubles as a coarse-grained human activity recognition frontend.Experimental results, using commercial micro-kinetic generators, demonstrate the flexibility and potential of this approach: the circuit achieves a quiescent current of 57 nA and a maximum load current of 300 mA, delivered with a harvesting efficiency of 79 %. Based on empirically collected motion data, the system achieves an energy surplus of over 232 mJ per day in a wrist-worn application while executing activity recognition at an accuracy of 89 % and a latency of 60 s.

Enhanced charge circuitry for indoor photovoltaic energy harvesting with fast activation and high efficiency

“…It is commonly utilized in energy harvesting applications such as EH, solar chargers, thermal electric generators (TEG), wireless sensor networks (WSN), and portable and wearable health devices. The TI BQ25570 chip is readily available in the market, and its features have been extensively discussed in numerous papers [14][15][16].…”

Exploring Artificial Neural Networks Efficiency in Tiny Wearable Devices for Human Activity Recognition

“…Recently, Alessandrini et al, presented a recurrent neural network (RNN), deployed on an embedded device, which takes in input data from Photoplethysmography (PPG) and tri-axial accelerometer sensors to infer the current human activity [ 30 ]. Similarly, Coelho et al [ 17 ] and Mayer et al [ 31 ] showed the adequacy of different deep learning models to be run on low-power platforms. Although the HAR on top of low-power devices is the common thread of these recent works, in none of the cases do authors analyze trade-offs between network complexity and resources utilization nor the impact on inference time or on measured energy consumption in real conditions.…”

Model-based design for self-sustainable sensor nodes

“…Matching is particularly crucial for small-sized harvesters with an output power of a view microwatts that need to supply wireless sensor nodes with peak power of milliwatts. Finally, there is not only the requirement to study and optimize the circuitry itself but also to precisely analyze the application-specific energy statistics to ensure long-term reliability [23]. A resilient power path design capable of overcoming temporary absences of environmental power, e.g., with cold start circuitry, allows to partially circumvent this critical design point but does not ensure continuous availability [24].…”