Radar State Detection for Accurate Fall Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems inaccurately identify a person's state, leading to unnecessary power and network resource consumption and false alerts, particularly in fall detection scenarios.
Innovation Solution
An active reflected wave detector, such as a radar sensor, is used to accurately classify a person's state by analyzing reflected wave measurements, including Doppler values and spatial distributions, and employing a trained classifier model to distinguish between standing, falling, and other states.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If reflected-wave based systems (radar, lidar, sonar) are used to monitor a person in a designated space, then the system can detect presence and position without contact, but the system may inaccurately identify the person's state leading to false alerts
Solution Approach 1:
The patent segments the reflected wave measurements into multiple components: Doppler values (velocity information) and spatial distributions (position information). By analyzing these segments separately and combining them through the classifier, the system achieves more accurate state classification while maintaining non-contact monitoring capability.
Solution Approach 2:
The patent changes the parameters used for state detection from simple presence/position detection to analyzing Doppler values and spatial distribution patterns. This parameter transformation enables the system to distinguish between different states (standing, falling, sitting) more accurately, reducing false alerts while maintaining ease of operation.
2Reliability
If fall detection systems transmit alert signals to remote devices, then the system can provide monitoring and response capability, but power and network resources are consumed unnecessarily when false alerts occur
Solution Approach 1:
The patent performs preliminary state classification using the trained classifier model before triggering any alert signals. By pre-analyzing the reflected wave measurements (Doppler values and spatial distributions) to determine the person's state, the system can prevent unnecessary alert transmissions, thereby conserving power and network resources while maintaining reliable fall detection capability.
3Measurement precision
If the system continuously monitors and classifies person states, then accurate state identification is achieved, but computational resources and processing time are increased
Solution Approach 1:
The patent uses a trained classifier model that has been pre-trained on reflected wave measurement patterns. Instead of performing complex real-time analysis from scratch, the system applies the pre-trained classifier to new measurements, significantly reducing computational requirements while maintaining high state classification accuracy.
Solution Approach 2:
The patent extracts the most informative features from the reflected wave measurements - specifically Doppler values and spatial distribution patterns - and feeds only these extracted features to the classifier. This feature extraction approach reduces the computational burden compared to processing all raw data, while still achieving accurate state identification.
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
Accurately identifies a person's state, reducing false alarms and conserving resources by differentiating between safe and hazardous conditions, thereby optimizing response efficiency.
Implementation Method 1
an active reflected wave detector, such as a radar sensor
Implementation Method 2
analyzing reflected wave measurements
Implementation Method 3
including Doppler values and spatial distributions
Data Source
Figure 1
Figure 2
Figure 3(a)~3(b)
AI summary
Embodiments relate to using an active reflected wave detector (206) to classify the state of a person (106) in an environment (100) and optionally respond accordingly. In one embodiment there is provided a computer implemented method (500) of determining a state of a person comprising: receiving (S502) an output of an active reflected wave detector; classifying (S508) a state of the person as being in a safe supported state based on the output using measurements of reflections associated with the person, wherein said classifying is based at least on: a height metric associated with at least one reflection from the person conveyed in the output of the active reflected wave detector; and a plurality of velocity magnitude measurements of the person corresponding to different times, each of said velocity magnitude measurements determined using the reflections associated with the person conveyed in the output of the active reflected wave detector.