Parking Assistance Using Learned Object References in Low-GPS Areas
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing parking assistance systems face challenges in accurately determining the proximity of a registered target parking position when the detection precision of the vehicle's self-position is low, such as in indoor or poorly mapped areas.
Innovation Solution
The system stores learned target object data in association with a first self-position detected before precision drops, allowing retrieval of this data when precision is low, using odometry and feature matching to determine proximity to the target parking position.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the system relies on real-time self-position detection, then the parking assistance can operate in real-time, but the determination accuracy fails when detection precision is low
Solution Approach 1:
The system performs preliminary actions by detecting and storing target object data and first self-position information when detection precision is high, before entering low-precision areas. This pre-stored data is then used for proximity determination when the vehicle is parked in low-precision environments, eliminating the need for accurate real-time positioning in those areas.
Solution Approach 2:
The system creates a copy of the target parking position information by storing target object data (images, features) and associated first self-position information in a storage device. This copied data serves as a reference template that can be compared against current sensor data to determine proximity without requiring high-precision real-time positioning.
2Adaptability or versatility
If the system operates in low-precision environments, then the parking assistance becomes more versatile, but the ability to determine proximity accurately deteriorates
Solution Approach 1:
The system prepares reference data (target object data and first self-position) in advance when in high-precision environments, enabling it to operate effectively later in low-precision environments by comparing current sensor inputs against this pre-stored reference information.
Solution Approach 2:
The system introduces target object data (images and features of surrounding objects) as an intermediary reference. Instead of directly relying on imprecise self-position data, the system uses these stored object features as a mediator to indirectly determine proximity by comparing current sensor data with the stored reference objects.
3Measurement precision
If the system stores position data only when precision is high, then data quality improves, but the system cannot assist parking when precision drops below threshold
Solution Approach 1:
The system performs the data storage action preliminarily - it detects and stores target object data and first self-position information in advance when detection precision meets the threshold. This ensures high-quality reference data is available before the system needs to operate in low-precision areas, maintaining both data quality and system availability.
Solution Approach 2:
The system serves itself by automatically detecting when detection precision is sufficient and autonomously storing the necessary reference data without external intervention. This self-service mechanism ensures that the system always has the required reference information available when needed, regardless of subsequent changes in detection precision.
Data Source
AI summary
A parking assistance method includes: setting a self-position at a point where the detection precision of the self-position being greater than or equal to a predetermined precision changes to less than the predetermined precision, as first self-position; detecting a relative positional relationship between a target object existing in surroundings of a target parking position when the own vehicle is parked at the target parking position after the first self-position is set and the target parking position; when the detection precision is less than a predetermined precision when the own vehicle stops at the target parking position, storing the first self-position and learned target object data representing the relative positional relationship, in association with each other in a storage device; and when the first self-position and the learned target object data are stored in association with each other, assisting parking based on the first self-position and the learned target object data.