Human activity recognition using magnetic induction-based motion signals and deep recurrent neural networks
Inventors
GOLESTANI, Negar • Moghaddam, Mahta
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12383162-B2
Publication Date
2025-08-12
Expiration Date
Abstract
A human activity recognition system includes a receiver coil mounted on a subject and a plurality of transmitter coils mounted on the subject at different locations than the receiver coil and each other. Each transmitter coil is inductively coupled to the receiver coil. A data processing system is in electrical communication with the receiver coil. Characteristically, the data processing system receives signals from each transmitter coils wherein the data processing system applies a machine learning classifier to determine an activity that the subject is engaged in.
Core Innovation
The invention relates to a human activity recognition system that uses magnetic-induction coupling between a receiver coil and a plurality of transmitter coils mounted on a subject. The receiver coil is mounted on the subject and inductively coupled to the transmitter coils via a non-propagating magnetic field so that the receiver coil receives signals from the plurality of transmitter coils.
A data processing system receives signals from each transmitter coil and applies a machine learning classifier to determine an activity engaged in by the subject. The classifier is trained to classify received signals into predetermined activity classifications including walking, running, jumping, knocking, lifting, and throwing, and the received signals are associated with a combination of the receiver coil and each transmitter coil having an operating frequency from 1 to 30 MHz.
To support multiplexing and robust signal acquisition from multiple inductive systems, multiplexing or frequency splitting is applied to reduce or eliminate interference between inductive systems. Each of the receiver coil and each transmitter coil is independently attached to a corresponding matching network for each coil, and data collection and wireless signal transmission are combined.
Claims Coverage
The document includes two independent claims: one system claim and one method claim. Both independent claims share the same core inventive concept and include six inventive features.
Non-propagating magnetic-field inductive coupling between a receiver coil and multiple transmitter coils
A receiver coil mounted on a subject and a plurality of transmitter coils mounted on the subject at different locations than the receiver coil and each other, wherein each transmitter coil is inductively coupled to the receiver coil such that the receiver coil receives signals from the plurality of transmitter coils, and wherein the transmitter coils are coupled to the receiver coil with a non-propagating magnetic field.
Machine learning classification of predetermined activities from received multi-coil signals
A data processing system in electrical communication with the receiver coil, receiving signals from each of the transmitter coils, applying a machine learning classifier to determine an activity engaged in by the subject, and trained to classify received signals into predetermined activity classifications including walking, running, jumping, knocking, lifting, and throwing.
Multiplexing or frequency splitting to reduce or eliminate interference between inductive systems
Multiplexing or frequency splitting is applied to reduce or eliminate interference between inductive systems.
Independently attached matching networks for receiver coil and each transmitter coil
Each of the receiver coil and each transmitter coil in the plurality of transmitter coils are independently attached to a corresponding matching network for each coil.
Operating frequency range for the receiver-transmitter coil combination
A combination of the receiver coil and each transmitter coil has an operating frequency from 1 to 30 MHz.
Combining data collection with wireless signal transmission
Data collection and wireless signal transmission are combined.
Across the independent system and method claims, the inventive coverage centers on inductively coupling multiple transmitter coils to a receiver coil via a non-propagating magnetic field, classifying received signals with a machine learning classifier into predetermined activities, and reducing interference using multiplexing or frequency splitting, supported by independently attached matching networks per coil and an operating frequency range of 1 to 30 MHz.
Stated Advantages
Reduced power/bandwidth needs.
Lower SAR in lossy media.
Improved QoS vs Bluetooth/radio-wave propagation.
Documented Applications
Human activity recognition using an MI-HAR system to classify activities including walking, running, jumping, knocking, lifting, and throwing.
Interested in licensing this patent?