Wi-Fi CSI Human Pose Detection via RF Sensing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing human pose estimation technologies face challenges such as occlusion and lighting issues with image-based methods and require expensive, power-intensive hardware like LiDAR and radar, raising privacy concerns when used in outdoor monitored sites like oil facilities.
Innovation Solution
The use of Wi-Fi emissions and advanced signal processing techniques to detect human presence and pose using Wi-Fi transceivers, converting channel state information into image representation maps and surface maps for accurate human activity monitoring without visual data, leveraging machine learning networks and existing infrastructure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image-based technologies (RGB cameras) are used for human pose estimation, then human presence and pose can be detected, but the system is adversely affected by occlusion and lighting issues
Solution Approach 1:
The patent replaces image-based optical sensing with radio frequency (RF) sensing technology. Instead of using cameras that capture visual information, the system employs RF signals to detect human presence and pose by measuring changes in the electromagnetic environment caused by human bodies. This substitution eliminates the fundamental limitations of visual systems regarding occlusion and lighting conditions, as RF waves can penetrate through obstacles and are unaffected by light levels.
Solution Approach 2:
The system changes the detection parameter from visual spectrum data to radio frequency signal characteristics. By monitoring RF signal phase, amplitude, and time-of-flight variations, the system extracts human presence and pose information without relying on visual parameters. This parameter transformation allows detection to proceed independently of lighting conditions and occlusion states that plague image-based methods.
2Measurement precision
If LiDAR and radar-based technologies are used for human detection, then detection accuracy can be improved, but specialized hardware is required that is expensive and power-intensive
Solution Approach 1:
The patent leverages the ubiquity of Wi-Fi transceivers and RF sensing capabilities already present in common wireless communication devices. Instead of requiring specialized LiDAR or radar hardware, the system utilizes the RF transceivers that are already widely deployed for wireless networking. This multi-functional approach allows the same hardware to serve both communication and sensing purposes, dramatically reducing system cost and complexity while maintaining effective human detection capabilities.
Solution Approach 2:
The system employs RF signals that are already being transmitted for wireless communication purposes. By repurposing these existing communication signals for dual use in both data transmission and human detection, the system eliminates the need for separate dedicated sensing hardware. The Wi-Fi transceivers serve themselves double duty, providing both network connectivity and sensing functionality without requiring additional specialized components.
3Measurement precision
If LiDAR and radar sensors are placed in non-public areas for monitoring, then human activity can be detected, but significant privacy concerns are raised
Solution Approach 1:
The patent introduces RF signal measurements as an intermediary medium for detection. Instead of directly capturing visual images that reveal detailed personal information, the system uses RF signal variations as an indirect mediator to infer human presence and general pose. This intermediary approach provides a privacy-preserving mechanism where detection information is obtained without direct visual observation, reducing the privacy intrusion while maintaining detection effectiveness.
Solution Approach 2:
The system substitutes visual imaging systems with RF sensing technology. By replacing cameras and visual sensors with radio frequency detectors, the system fundamentally changes the detection mechanism from direct visual capture to indirect electromagnetic field measurement. This substitution inherently reduces privacy concerns as RF sensing does not capture visual information that could reveal sensitive personal details, while still providing effective human presence and pose detection capabilities.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Provides real-time, non-intrusive, and reliable human activity detection, integrating with existing monitoring systems to enhance security and operational efficiency while avoiding the need for specialized hardware.
Implementation Method 1
leveraging Wi-Fi signals to detect anomalies in signal amplitude and shifts
Implementation Method 2
The TX emits Wi-Fi signals, and the RX receives and analyzes the signals
Implementation Method 3
converting channel state information into image representation maps, by performing domain translation of the CSI data using a two-branch encoder-decoder network
Implementation Method 4
generating, by a second ML network, a surface map based on the image representation. The surface map is a representation of correspondence between the image representation map and a 3-dimensional (3D) surface associated with a human body
Data Source
AI summary
Methods and systems for detecting human presence in an outdoor monitored site are disclosed. In an configuration, the method may include providing at least three transmitters for transmitting radio signals and at least three receivers for receiving radio signals. Channel state information (CSI) data is extracted from the radio signals. The CSI data include phase values and amplitude values. The CSI data is converted into an image representation map and then a surface map is generated using Machine Learning (ML) networks. The surface map is a representation of correspondence between the image representation map and a 3-dimensional (3D) surface associated with a human body.