Map-Based Trajectory Localization Using GNN and RNN Motion Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing localization technologies, such as GPS and Indoor Positioning Systems, face challenges in accuracy and reliability, especially in environments where satellite signals are degraded or unavailable, leading to high maintenance costs and inefficiencies.
Innovation Solution
A method utilizing a neural network to map motion trajectories on geospatial maps, employing a topological map-based approach with sensor fusion and neural networks like RNN and GNN to learn object motion patterns, enabling accurate localization through relative position signals and trajectory analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If GPS technology is used for outdoor positioning, then coverage area is improved, but measurement precision deteriorates due to signal degradation and multi-path problems
Solution Approach 1:
The patent introduces map data and visual features as intermediary elements between the camera and the final position estimate. Instead of directly using GPS coordinates, the system captures images, extracts visual features, matches them against pre-stored map data, and derives position information from the map structure. This intermediary approach allows the system to achieve accurate positioning in environments where GPS signals are degraded or unavailable.
2Measurement precision
If Indoor Positioning Systems with WiFi, Bluetooth, or UWB infrastructure are deployed, then measurement precision is improved, but device complexity and maintenance cost increase
Solution Approach 1:
The system uses the device's existing camera and processor to perform positioning tasks without requiring external positioning infrastructure. The mobile device itself captures images, processes visual features, and determines position using algorithms running locally. This self-service approach eliminates the need for deployed WiFi, Bluetooth, or UWB infrastructure, reducing both device complexity and maintenance costs while maintaining positioning accuracy.
3Reliability
If traditional localization methods are used in GPS-denied environments, then reliability deteriorates, but implementing alternative systems increases loss of time and computational resources
Solution Approach 1:
The patent pre-stores map data including visual features, geometric structures, and position information before the actual positioning task. By having this reference data ready in advance, the system can quickly match captured images against the pre-stored map without performing complex computations in real-time. This preliminary preparation significantly reduces processing time during actual positioning while maintaining high reliability in GPS-denied environments.
Data Source
AI summary
An object of initial unknown position on a map may be determined by traversing through moving and turning to establish motion trajectory to reduce its spatial uncertainty to a single location that would fit only to a certain map trajectory. A artificial neural network model learns from object motion on different map topologies may establish the object's end-to-end positioning from embedding map topologies and object motion. The proposed method includes learning potential motion patterns from the map and perform trajectory classification in the map's edge-space. Two different trajectory representations, namely angle representation and augmented angle representation (incorporates distance traversed) are considered and both a Graph Neural Network and an RNN are trained from the map for each representation to compare their performances. The results from the actual visual-inertial odometry have shown that the proposed approach is able to learn the map and localize the object based on its motion trajectories.