Machine Learning Wave Property Estimation Using IMU Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Underwater devices, such as underwater camera systems, face challenges due to high-intensity tidal waves, which can cause damage or impact their operation. Real-time information on wave properties is needed for safe deployment and operation of these devices.
Innovation Solution
A machine learning model processes measurement data from an onboard inertial-measurement unit (IMU) to estimate wave properties such as average wave height, wave frequency, wave energy, and wave power density distribution. This data is used to determine whether the device is safe to be deployed and to control its position accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If real-time wave property measurement is implemented using traditional sensors, then wave property information can be obtained, but the device complexity and cost increase significantly
Solution Approach 1:
The patent reuses the existing IMU device, which was originally designed for other purposes (such as navigation or stabilization), and enables it to also measure wave properties. This multi-functional approach allows wave height, wave period, and other wave parameters to be derived from the same sensor that provides orientation and position data, thereby avoiding additional specialized wave measurement equipment.
Solution Approach 2:
The system uses the underwater device's own IMU sensor to measure wave properties rather than requiring separate dedicated wave measurement instruments. The device serves itself by utilizing its existing onboard sensors and processing capabilities to characterize the wave environment, eliminating the need for external measurement systems.
2Measurement precision
If dedicated wave measurement equipment is deployed, then accurate wave property data is obtained, but the deployment complexity and risk increase
Solution Approach 1:
The existing IMU device performs both its original function (such as navigation or stabilization) and wave measurement functions simultaneously. This eliminates the need for separate dedicated wave measurement equipment that would require additional deployment procedures, reducing operational complexity while maintaining measurement capability.
Solution Approach 2:
The patent combines wave measurement functionality with the existing IMU system. By integrating wave property extraction algorithms into the IMU data processing pipeline, the system merges multiple functions (positioning, orientation, and wave measurement) into a single unified system, simplifying deployment and reducing the number of separate components required.
3Loss of information
If traditional wave measurement methods are used, then wave properties can be characterized, but the real-time processing capability is insufficient
Solution Approach 1:
The patent replaces traditional mechanical wave measurement systems (such as wave buoys or optical measurement devices) with a computational approach using machine learning models. The IMU data is processed through trained neural networks or other ML algorithms that rapidly estimate wave properties in real-time, providing faster processing compared to traditional mechanical measurement and analysis methods.
Solution Approach 2:
The system transforms the approach to wave measurement by changing from direct physical measurement to indirect computational estimation. By using machine learning models trained on wave data, the system transforms IMU sensor readings into wave property estimates through parameter transformation, enabling real-time processing with lower computational overhead than traditional methods.
Data Source
AI summary
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for estimating wave properties of a body of water. A computer-implemented system obtains measurement data for a duration of time from an inertial measurement unit (IMU) onboard an underwater device, generates model input data based on at least the measurement data obtained at the plurality of time points, and processes the model input data to generate model output data indicating one or more wave properties using a machine-learning model. The system further determines, based on at least the one or more wave properties, whether the device is safe to be deployed.