UWB Key Positioning via Virtual Environment Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining the position of a key device relative to a vehicle using smartphones as key devices face challenges due to the complex electromagnetic behavior of vehicles and the need for extensive training data across various environments, especially with ultra-wideband (UWB) signals which are heavily influenced by environmental reflections.
Innovation Solution
A machine learning model is trained using data from two distinct vehicle environments that differ significantly in terms of reflections, allowing it to model behavior in intermediate environments without requiring additional measurements, by utilizing time-of-flight distance measurements and signal strength data from UWB signals, and augmenting data through simulation and noise addition to cover a broader solution space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ultra-wideband signals are used for distance measurement in smartphones, then measurement precision is improved, but the system becomes more sensitive to environmental reflections and interference
Solution Approach 1:
The patent creates virtual training environments that replicate real-world electromagnetic conditions. Instead of physically testing in every possible environment, the system generates synthetic training data that copies the characteristics of different environments (garages, open spaces, urban areas), allowing the machine learning model to learn reflection patterns without being exposed to actual environmental interference during training.
Solution Approach 2:
The system varies electromagnetic parameters in the training data, including signal frequency, power levels, and environmental reflection coefficients. By changing these parameters across multiple virtual training scenarios, the model learns to distinguish between direct signals and reflected signals, maintaining measurement precision while becoming robust to environmental variations.
2Reliability
If training data is collected from multiple real-world environments, then reliability is improved, but training effort and complexity increase significantly
Solution Approach 1:
The patent replaces physical collection of training data from multiple real environments with virtual copying of these environments through simulation. The system creates digital twins of different parking environments (garages with concrete walls, open spaces, urban areas) and generates training data by simulating UWB signal propagation in these virtual environments, dramatically reducing the time and effort required compared to physical data collection.
Solution Approach 2:
The system performs preliminary action by pre-generating comprehensive training datasets that cover all possible environmental scenarios before the actual classification task. The virtual training environments are set up in advance with pre-calculated electromagnetic propagation characteristics, so when the model needs to classify a key's position, the training is already complete and the model is ready for immediate use.
3Reliability
If the solution space is fully covered with training data, then classification reliability is improved, but the complexity of data collection and processing increases
Solution Approach 1:
The patent adds a virtual dimension to the training process by creating synthetic environments that extend beyond physical limitations. Instead of collecting data only from physically accessible locations, the system generates training data in virtual 3D spaces that can represent any possible environment configuration, covering the entire solution space without the complexity of physical data collection from every location.
Solution Approach 2:
The virtual training environment system serves multiple functions simultaneously: it generates training data for different environments, simulates various electromagnetic conditions, tests model performance, and validates classification algorithms all within a single unified platform. This multi-functionality reduces the overall complexity compared to separate systems for each function.
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
This approach reduces the training effort and enhances the reliability of determining whether a key device is inside or outside the vehicle, effectively handling diverse vehicle environments by leveraging machine learning to model reflection behaviors across different scenarios.
Implementation Method 1
The use of a very broad spectrum (of at least 500 MHz, for example) makes it possible to carry out a precise time-of-flight (ToF) measurement. The ToF measurement can then be used to calculate the distance between the transmitter and the receiver using the constant of the speed of light.
Implementation Method 2
a defined signal is emitted by the key and one or more receiving antennas receive this signal with different signal strengths depending on the position in relation to the vehicle. The reason for this is the attenuation of an electromagnetic wave in different materials.
Implementation Method 3
At conductive geometries of the order of magnitude of the wavelength, an electromagnetic wave interacts very strongly, that is to say the wave is reflected, scattered and diffracted.
Implementation Method 4
At conductive geometries of the order of magnitude of the wavelength, an electromagnetic wave interacts very strongly, that is to say the wave is reflected, scattered and diffracted.
Implementation Method 5
At conductive geometries of the order of magnitude of the wavelength, an electromagnetic wave interacts very strongly, that is to say the wave is reflected, scattered and diffracted.
Data Source
AI summary
A computer-implemented method trains a machine learning model. The method includes training the machine learning model on the basis of data representing at least two different vehicle environments. The machine learning model is trained, on the basis of data from a time-of-flight distance measurement of a distance between a key device and a vehicle, to determine a position of the key device relative to the vehicle.