Researchers at Rutgers University have developed a battery-free AI wearable embedded in a shoe sole that analyzes a person's gait in real time, powered entirely by energy harvested from walking.
Described in a peer-reviewed paper in Science Advances, the device combines a biomechanical energy harvester, accelerometer, low-power edge-AI computing, and cold-start power management into a self-sustaining platform. The system consumes just 86 microwatts and distinguished four locomotion modes with 95.4% accuracy.
The key innovation: the shoe generates enough energy from each step to power sensing and analysis, eliminating the need for a rechargeable battery. A multilayer energy harvester captures the repetitive motion of walking, while the edge-AI sensor processes accelerometer data locally rather than transmitting it to a phone or cloud.
While the prototype was tested on only four healthy volunteers and is not yet a clinical device, it addresses a fundamental limitation of current wearables — the need for regular charging. The researchers describe the design as biomimetic, inspired by organisms that sustain themselves from environmental energy.
