Machine-Learning Motion Tracking With Secondary Sensors During GNSS Loss
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inertial navigation systems using low-cost IMUs suffer from significant error drift due to sensor noise, biases, and alignment errors, leading to inaccurate positioning and navigation, especially in situations where GNSS data is unavailable or unreliable.
Innovation Solution
A method utilizing a trained machine learning algorithm to estimate a metric of interest using data from secondary sensors mounted on separate platforms, such as wearable devices, to constrain tracking solutions when primary positioning units like GNSS are not operational.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If low-cost IMUs are used in inertial navigation systems, then device cost and accessibility are reduced, but measurement precision and reliability deteriorate due to sensor noise, biases, and alignment errors
Solution Approach 1:
The system performs preliminary calibration of the low-cost IMU sensors during manufacturing or initial setup to characterize and compensate for sensor biases, scale factors, and alignment errors. This pre-characterization allows the system to achieve acceptable navigation accuracy without requiring expensive high-precision sensors.
Solution Approach 2:
The system employs feedback mechanisms where navigation solutions from multiple sensors (including low-cost IMUs) are continuously integrated and corrected. Error drift from individual low-cost sensors is compensated through feedback from other sensors and algorithmic correction, maintaining overall system accuracy.
2Productivity
If numerical integration is used to determine velocity and position from accelerometer and gyroscope measurements, then tracking and navigation data can be derived, but error drift increases rapidly over time
Solution Approach 1:
The system merges data from multiple independent sensors (accelerometers, gyroscopes, magnetometers, barometers, GPS receivers) to determine navigation parameters. By combining these different measurement sources, the system cross-validates results and compensates for the cumulative error drift that would occur with numerical integration alone.
Solution Approach 2:
The navigation system is designed to perform multiple functions simultaneously: it determines orientation from gyroscopes, velocity from accelerometers, position from GPS and barometric pressure, and heading from magnetometers. This multi-functional approach allows error compensation across different measurement domains, reducing overall error drift.
3Measurement precision
If GNSS receivers are used in combination with IMUs to improve navigation accuracy, then positioning precision is enhanced, but the system fails in urban canyons or indoor environments where satellite visibility is blocked
Solution Approach 1:
The system integrates multiple positioning methods (GPS satellite-based positioning, visual odometry using cameras, inertial navigation using IMUs, and barometric altimetry) into a single universal navigation solution. This allows the system to automatically switch between or combine different positioning techniques depending on environmental conditions, maintaining functionality whether GPS is available or blocked.
Solution Approach 2:
Visual odometry using camera-based feature tracking serves as an intermediary positioning method between GPS and pure inertial navigation. When GPS signals are blocked in urban canyons or indoors, the camera-based visual odometry provides continuous positioning by tracking visual features in the environment, bridging the gap where satellite-based positioning fails.
4Ease of operation
If smart devices are carried in various positions and orientations with respect to the user, then ease of operation is improved, but measurement precision deteriorates due to changing motion characteristics
Solution Approach 1:
The system uses data from multiple sensors (accelerometers, gyroscopes, magnetometers) to determine device orientation and compensate for its position relative to the user's center of mass. By fusing these measurements, the system can accurately track user motion regardless of how the device is held or worn, maintaining measurement precision while allowing flexible device placement.
Solution Approach 2:
The system dynamically adjusts its measurement and processing parameters based on detected device orientation and motion context. By changing parameters such as integration methods, sensor weighting, and coordinate transformation approaches according to the device's actual state, the system maintains accuracy across varying device positions and orientations.
Data Source
AI summary
There is disclosed a computer-implemented method performed in a tracking system for tracking the motion of a body, as a function of time, the method comprising: (a) during a first time period, obtaining first data related to the motion of a body from at least one primary positioning unit, wherein said at least one primary positioning unit is mounted on a first platform carried on the body, or wherein said at least one primary positioning unit is separate to the body, said primary positioning unit being operational during the first time period; (b) during the first time period, obtaining second data from one or more secondary sensors configured to make measurements from which position or movement may be determined, said one or more secondary sensors being mounted on one or more second platforms carried on the body; (c) generating first training data comprising the first data and second data; (d) during a second time period, obtaining third data from the one or more secondary sensors, and; (e) analysing the third data to estimate at least one first metric related to the motion of the body during the second time period using a first algorithm trained using the first training data. A tracking system for tracking the motion of a body, as a function of time, is also disclosed.