3D Vision-GNSS Fusion for Urban Canyon RTK Under NLOS Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing GNSS-RTK positioning techniques suffer from degraded accuracy in urban canyons due to NLOS receptions and distorted satellite geometry, which are not effectively addressed by current methods that rely on expensive sensors like 3D LiDAR or require accurate 3D building models.
Innovation Solution
A method that integrates low-lying visual landmarks from a forward-looking camera with high-elevation GNSS measurements using a tightly coupled factor graph optimization, incorporating IMU pre-integration and Doppler measurements to improve geometry constraints and resolve integer ambiguities, without relying on 3D LiDAR.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GNSS-RTK positioning is used in urban canyons, then centimeter-level positioning accuracy can be achieved in open areas, but the accuracy is significantly degraded due to NLOS receptions and distorted satellite geometry
Solution Approach 1:
The patent segments the satellite measurements into two categories: high-elevation satellites (above a threshold angle) and low-elevation satellites. The factor graph optimization selectively processes these segments differently, using high-elevation satellites for reliable positioning while identifying and excluding NLOS receptions from low-elevation satellites, thereby resolving the contradiction between utilizing sufficient satellite geometry and avoiding harmful NLOS signals
Solution Approach 2:
The patent introduces a factor graph optimization as an intermediary computational framework that mediates between raw GNSS measurements and final position estimates. This intermediary layer applies geometric constraints and evaluates measurement reliability, filtering out NLOS receptions while preserving valid low-elevation satellite data, thus maintaining positioning accuracy in urban canyon environments
2Reliability
If 3D LiDAR is used for positioning, then robustness and accuracy are improved, but the cost increases significantly preventing massive deployment
Solution Approach 1:
The patent replaces expensive 3D LiDAR sensors with a combination of low-cost forward-looking cameras and standard GNSS receivers. The camera captures images for visual landmark detection, while the GNSS receiver collects satellite measurements. This substitution uses inexpensive, widely available components to achieve positioning reliability previously only attainable with costly 3D LiDAR systems
Solution Approach 2:
The patent substitutes the mechanical 3D LiDAR scanning system with an optical camera-based visual landmark recognition system combined with electromagnetic GNSS signal processing. The camera captures visual features of low-lying landmarks, and the factor graph optimization integrates these visual constraints with GNSS measurements, replacing the mechanical ranging approach of LiDAR with an optical-electromagnetic hybrid approach that is far more cost-effective
3Device complexity
If VINS is used for navigation, then size, power assumption, weight, and availability are improved, but the state estimation is subject to drift over time
Solution Approach 1:
The patent incorporates global feedback from GNSS-RTK measurements into the visual-inertial navigation system. The factor graph optimization continuously integrates GNSS position estimates with VINS state estimates, using the globally referenced GNSS data to correct and reset the cumulative drift inherent in inertial and visual odometry, thereby maintaining long-term positioning stability while preserving the compact and low-power advantages of VINS
Solution Approach 2:
The patent merges the local relative positioning capabilities of VINS with the global absolute positioning capabilities of GNSS-RTK into a unified factor graph optimization framework. This combination allows the system to benefit from both approaches: the compact, low-power VINS architecture for continuous navigation and the globally referenced GNSS measurements for drift correction, achieving stable long-term positioning without sacrificing system simplicity
Data Source
AI summary
In estimating a position of a vehicle utilizing a global navigation satellite system (GNSS), it is desirable to exclude outliner GNSS measurements due to navigation signals via non-light-of-sight paths from the GNSS to the vehicle, but it leads to a distorted satellite geometry distribution. Complementariness between low-lying visual landmarks and healthy but high-elevation satellite measurements is explored to improve the geometry constraint. Measurements of an inertial measurement unit, low-lying visual landmarks captured by a forward-looking camera onboard the vehicle, and healthy but high-elevation satellite measurements are tightly-coupled integrated via sliding window optimization of system states used in a factor graph. To improve estimation performance, good initial guesses of system states are important. As such, initial guesses of velocity set and position set inside a sliding window are estimated simultaneously based on data of Doppler measurement, double-differenced (DD) pseudorange measurement and DD carrier-phase measurement as obtained in GNSS measurements.