Radar-based Motion Classification Using One or More Time Series

Simple SummaryContent extracted from patent full text and abstract with AI.

This patent describes a computer-implemented method and system for classifying motions, such as hand gestures, using radar measurements. By capturing sequences of radar data and extracting one or more one-dimensional time series for different motion-related observables (like range, amplitude, azimuth, elevation), the invention employs machine learning algorithms—such as convolutional or autoencoder neural networks—to quickly and accurately identify specific gestures or motion classes. Unlike conventional camera-based systems, it is robust to lighting, occlusions, and protects user privacy.

Use CasesContent extracted from patent full text and abstract with AI.

• Touchless control of smart devices (phones, TVs, appliances) via hand or body gestures
• Gesture-based control interfaces in vehicles, e.g., infotainment or trunk opening via kick gesture
• Hands-free user interaction with vending, ticketing, or public kiosks
• Gesture recognition in virtual and augmented reality (AR/VR) systems
• Sign language interpretation interfaces
• Monitoring and recognizing body poses or facial expressions for accessibility or gaming
• Smart home automation by recognizing gestures to control lights, alarms, or multimedia systems
• Adaptive control of automotive door handles (e.g. extending door handles upon gesture)

BenefitsContent extracted from patent full text and abstract with AI.

• Works in varying lighting conditions and is immune to visual occlusion issues that affect camera-based systems
• Reduces privacy concerns since radar does not capture images of users
• Fast and lightweight computational requirements, suitable for embedded or edge devices
• Highly accurate recognition rates for complex and subtle gestures
• Can be implemented on standard hardware without the need for specialized digital signal processors
• Easily extensible to recognize new, undefined gestures or motion classes
• Offers touchless interaction, enhancing hygiene and convenience
• Compact data representation enables low power consumption and faster reaction times

Technical Classifications (CPCs)

Inventors & Applicants

Patent Abstract

Techniques are disclosed that facilitate motion classification, e.g., gesture classification, based on one or more one dimensional time series (101-104) that are obtained from a radar measurement.