Vehicle Map Feature Point Pruning for Accurate Position Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional map generation devices for vehicles struggle to efficiently remove unnecessary feature points from the map due to changes in road structures or environments, leading to inaccurate tracking and data redundancy.
Innovation Solution
A map generation apparatus that includes a feature point extraction unit, a position recognition unit, and a deletion unit, which extracts feature points from in-vehicle detection data, collates them with the map, and deletes points with low recognition results, thereby reducing unnecessary data and maintaining map accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If feature points are continuously accumulated in the map during vehicle travel, then the map coverage and feature point quantity increase, but unnecessary feature points that do not match the current road environment accumulate, reducing tracking accuracy
Solution Approach 1:
The patent implements a mechanism to discard unnecessary feature points from the map based on their usage frequency and relevance. The deletion unit removes feature points that have not been successfully matched or are no longer valid, while preserving useful feature points. This selective discarding maintains map accuracy by eliminating redundant data that accumulates during continuous vehicle travel and map updates.
Solution Approach 2:
The patent changes the state of feature points by tracking their collation results and match frequencies. Feature points are dynamically evaluated based on parameters such as the number of successful matches and collation accuracy. This parameter-based evaluation enables the system to identify and remove feature points that no longer meet the required accuracy thresholds, thereby maintaining reliable tracking despite continuous map accumulation.
2Loss of information
If all extracted feature points are retained in the map, then complete road environment information is preserved, but data redundancy increases and processing efficiency decreases
Solution Approach 1:
The patent systematically discards redundant feature points while preserving essential road environment information. The deletion unit evaluates each feature point's contribution to accurate positioning and removes those that are duplicate, outdated, or irrelevant. This process maintains information completeness for valid features while eliminating redundancy that hinders processing efficiency.
Solution Approach 2:
The patent extracts and removes specific unnecessary feature points from the map data structure. The deletion unit identifies and extracts redundant features based on collation results and match frequency, separating useful information from waste. This extraction process reduces data volume and improves processing efficiency without compromising the completeness of essential road environment information.
3Quantity of substance
If feature points are deleted based on lack of tracking, then data redundancy is reduced, but feature points that are temporarily undetected are incorrectly removed
Solution Approach 1:
The patent performs preliminary evaluation of feature points before deletion by tracking their collation results and match history over time. Rather than immediately deleting feature points that are temporarily undetected, the system accumulates evidence about their validity through repeated collation attempts. This preliminary action distinguishes between feature points that are temporarily invisible and those that are permanently invalid, preventing incorrect deletions.
Solution Approach 2:
The patent implements a feedback mechanism where collation results and match frequencies are continuously monitored and used to inform deletion decisions. The system receives feedback from the position recognition unit about which feature points are successfully matched and which are not. This feedback loop ensures that only feature points with consistently poor performance are removed, while temporarily undetected but valid features are retained.
Data Source
AI summary
A map generation apparatus includes: an in-vehicle detection unit configured to detect a situation around a subject vehicle in traveling; and a microprocessor and a memory connected to the microprocessor. The microprocessor is configured to perform: extracting one or more feature points from a detection data acquired by the in-vehicle detection unit; generating a map using the feature points extracted in the extracting while the subject vehicle is traveling; collating the feature points extracted in the extracting with the map to recognize a position of the subject vehicle on the map when the subject vehicle travels in a region corresponding to the map; and deleting, from the map, a feature point whose result of collation in the recognizing is less than a predetermined degree among the feature points included in the map.