GNSS Multi-Level Road Detection via Point Cloud Pattern Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current GNSS-based systems lack effective solutions for accurately determining positions on multi-level roads, which is critical for autonomous driving due to safety and integrity concerns regarding geolocation information.
Innovation Solution
A method involving the acquisition of actually tracked and theoretically trackable GNSS satellites, generating a point cloud representation, identifying obscured satellites using pattern recognition, and detecting road levels by identifying a roadway-like pattern in the point cloud, allowing for the determination of vehicle position on multi-level roads.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If GNSS signals are used for position determination, then navigation capability is provided, but on multi-level roads upper levels obscure satellite signals causing loss of tracking and reduced positioning accuracy
Solution Approach 1:
The patent transitions from 2D navigation maps to 3D point cloud representations. By generating a three-dimensional point cloud from GNSS satellite data and comparing it with expected satellite positions, the system can detect vertical obstructions (upper road levels) that block satellite signals. This dimensional upgrade enables the system to identify when a vehicle is on a lower level of a multi-level road by detecting the pattern of obscured satellites in the 3D space.
2Measurement precision
If 2D navigation maps are used, then map simplicity is maintained, but position accuracy on multi-level roads deteriorates due to lack of height information
Solution Approach 1:
The patent enhances traditional 2D navigation maps by integrating 3D point cloud data generated from GNSS satellite observations. The system creates a three-dimensional representation of the environment including vertical structures, enabling accurate determination of vehicle position on multi-level roads. This additional vertical dimension provides height information that resolves the ambiguity of position accuracy on elevated versus ground-level roads.
Solution Approach 2:
The system pre-generates point cloud representations and identifies roadway patterns (such as elevated road structures) before vehicle navigation. By having the 3D environmental model and expected satellite visibility patterns prepared in advance, the system can quickly compare actual satellite tracking data against predicted patterns to determine vehicle level on multi-level roads, rather than computing everything in real-time during navigation.
3Measurement precision
If diverse sensor technology is deployed, then sensing capability is enhanced, but solutions for recognizing positions on multi-level roads remain insufficient
Solution Approach 1:
The patent makes the GNSS receiver serve multiple functions: it not only determines basic position and velocity but also generates three-dimensional point cloud representations of the environment and detects vehicle level on multi-level roads. By utilizing the existing GNSS satellite signals and receiver capabilities in a novel way, the system achieves multi-level road position recognition without adding separate specialized sensors, thereby avoiding increased device complexity while improving measurement precision.
Data Source
AI summary
A method for detecting multi-level roads on the basis of GNSS includes (i) acquiring actually tracked GNSS satellites and the theoretically trackable GNSS satellites at a location, (ii) generating a point cloud representation using the actually tracked and the theoretically trackable GNSS satellites, (iii) discovering which theoretically trackable GNSS satellites cannot be tracked at the location, and (iv) detecting a road level as part of the detected positions when, in the point cloud representation, the discovered GNSS satellites are situated in a particular region corresponding to a roadway-like pattern.