Automatic Parking Feature Descriptor Map Adaptation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated parking systems require frequent retraining due to environmental changes, such as seasonal variations or sudden alterations like a knocked-down wall, which limits their operational lifetime and necessitates manual intervention.
Innovation Solution
The method involves manually driving a vehicle into a parking slot to create a digital map of environmental features, then automatically navigating it using feature descriptors, with repeated parking sessions counting matches and deleting descriptors below a threshold to dynamically update the map, allowing for adaptive learning and feature refinement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the system stores feature descriptors from a single training session, then the initial parking function is achieved, but the system requires frequent retraining when environmental changes occur
Solution Approach 1:
The system dynamically updates the digital map by comparing feature descriptors from multiple replay sessions against the stored training data. Feature descriptors that consistently match across sessions are retained, while those that diverge are marked for removal. This dynamic adaptation allows the system to maintain reliability without requiring manual retraining, as the map automatically adjusts to environmental changes through repeated use.
Solution Approach 2:
The system implements a feedback mechanism where each replay session provides information about environmental consistency. By counting matches between detected features and stored feature descriptors, the system receives feedback on which features remain stable and which have changed. This feedback loop enables automatic adjustment of the digital map, reducing the need for manual retraining while maintaining parking reliability.
2Adaptability or versatility
If the system performs multiple replay sessions to update the digital map, then adaptability to environmental changes improves, but system complexity increases
Solution Approach 1:
The system segments the feature descriptor validation process into distinct operations: detecting features in the environment, extracting feature descriptors, counting matches against the stored digital map, and selectively deleting mismatched descriptors. This segmentation allows the complex task of environmental adaptation to be broken into manageable steps that can be executed automatically during normal replay sessions, improving adaptability without proportionally increasing overall system complexity.
Solution Approach 2:
The system performs self-updating of the digital map by automatically comparing detected features with stored descriptors and removing inconsistent features without external intervention. This self-service capability allows the system to adapt to environmental changes autonomously through repeated replay sessions, enhancing versatility while keeping the user interface simple and the overall system architecture manageable.
3Measurement precision
If the system deletes feature descriptors below a threshold, then the digital map accuracy improves, but information loss occurs
Solution Approach 1:
The system changes the state of feature descriptors from static to dynamic by introducing a match-count parameter that tracks consistency across replay sessions. Feature descriptors are retained or deleted based on this parameter threshold, allowing the system to maintain high measurement precision by removing inconsistent features while preserving those that reliably represent the environment. This parameter-based approach minimizes information loss by retaining only the most stable and relevant features.
Data Source
AI summary
A method for automatically parking a vehicle in a parking slot involves manually driving the vehicle into the parking slot in a training step, and thereafter automatically driving the vehicle into the parking slot in a replay step. Automatically driving the vehicle into the parking slot involves detecting information of features of the environment of the vehicle corresponding to the driven trajectory, matching feature descriptors of the detected features of the environment with feature descriptors stored in the digital map, and re-localizing the vehicle against the trajectory stored in the digital map in order to navigate the vehicle along the stored trajectory into the parking slot. Automatically driving the vehicle into the parking slot is repeated multiple times, and involves deleting a feature descriptor stored in the digital map if the number of matches does not exceed a threshold after a predefined number of repetitions.