Vehicle Radar Localization Using Feature Map Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle localization methods using radar systems struggle in urban environments due to confusion between parked vehicles and recorded infrastructure, leading to self-localization failures, especially in lateral positioning.
Innovation Solution
A method that creates a feature map from successive radar scans to identify temporarily occupied regions by parked vehicles not recorded in the navigation map, applying a classifier to reduce the weight of radar detection points in these regions during matching, enhancing the robustness of the self-localization process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If Scan Matching algorithm is used to match radar detection points with navigation map elements, then vehicle position can be determined, but parked vehicles are confused with infrastructure elements leading to localization failure
Solution Approach 1:
The patent applies preliminary action by performing multiple radar scans over time before final matching. It collects radar detection points from successive scans, builds a feature map, and identifies temporarily occupied regions prior to the final position determination. This preliminary classification of objects as temporary or permanent prevents confused matching with infrastructure elements.
Solution Approach 2:
The patent implements dynamics by transitioning from static map matching to dynamic temporal analysis. It continuously updates the feature map with new radar scans, tracks object persistence over time, and adapts the matching process based on detected changes. This dynamic approach allows the system to distinguish between permanent infrastructure and temporary parked vehicles.
2Quantity of substance
If radar detection points from all regions are used equally in matching, then comprehensive data is utilized, but accuracy decreases due to inclusion of temporary objects not in navigation map
Solution Approach 1:
The patent applies local quality by differentiating the treatment of radar detection points based on their spatial and temporal characteristics. It creates a feature map with locally varying properties, identifying regions with temporary objects and assigning them different weights or exclusion status compared to regions with permanent infrastructure. This localized quality control improves matching precision.
Solution Approach 2:
The patent implements parameter changes by modifying the weight or significance parameter of radar detection points based on temporal analysis. Detection points in regions identified as temporarily occupied are downweighted or excluded from matching, while points in stable regions maintain full weight. This parameter adjustment optimizes the balance between data utilization and accuracy.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach improves the reliability and robustness of vehicle self-localization in urban environments by distinguishing between permanent and temporary objects, preventing ambiguities and ensuring accurate positioning even with multiple parked vehicles present.
Implementation Method 1
a radar system comprises at least one radar sensor (14) adapted to receive radar signals emitted from at least one radar emitter (14) of the radar system and reflected from an object in a vicinity of the vehicle towards the radar sensor
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for determining the position of a vehicle equipped with a radar system comprises successively acquiring radar scans respectively comprising a plurality of radar detection points, determining a predefined navigation map comprising at least one element representing a static landmark in the vicinity of the vehicle, matching at least a subset of the plurality of radar detection points of a radar scan and the at least one element of the navigation map, and determining the position of the vehicle based on the matching. Radar detection points of successive scans are collected over a time period and a feature map of the vicinity of the vehicle is created, wherein the feature map comprises a plurality of detection features of the collected radar detection points. A classifier is applied to the feature map to determine temporarily occupied regions which are likely to be occupied by temporarily present objects not recorded in the navigation map. Radar detection points which are located in a temporarily occupied region are considered to a lesser extent in the matching than radar detection points which are not located in a temporarily occupied region.