Garage Parking Space Detection Using Deep Learning Maps

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current garage parking systems often require manual activation by the driver due to the inability to accurately detect a garage parking space, leading to inefficiencies in semi-autonomous or fully automatic parking processes.

Innovation Solution

A method utilizing deep learning models with environmental sensor data, including ultrasonic, radar, and camera inputs, to create a digital map and classify a parking space as a garage space, enabling automatic detection and activation of the parking assistance system.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional sensor-based detection methods are used to identify parking spaces, then the system can detect basic parking space boundaries, but it cannot reliably distinguish garage parking spaces from regular parking spaces

Engineering Contradiction:
Improveparking space detection accuracyVSAvoidgarage identification reliability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The system transitions from basic geometric parameter detection to deep learning-based classification by changing the detection parameters to include semantic understanding of garage characteristics, enabling reliable distinction between garage and regular parking spaces

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent replaces traditional mechanical/sensor-based detection methods with deep learning model-based classification, substituting physical measurement systems with intelligent algorithms that can recognize garage parking spaces with high reliability

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Reliability

If the system requires manual activation by the driver, then it can avoid false activation, but it reduces automation efficiency and user convenience

Engineering Contradiction:
Improveactivation accuracyVSAvoidparking system automation level
Core Design Contradiction:
ReliabilityVSExtent of automation

Solution Approach 1:

The deep learning model enables the system to automatically identify and activate garage parking mode without driver intervention, making the system self-activating based on its own detection capabilities while maintaining high reliability through accurate garage recognition

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system uses feedback from the deep learning model's classification output to automatically determine when to activate garage parking mode, creating a closed-loop control system that balances automation with accurate detection

Inventive Principle:
Principle #23Feedback

3Measurement precision

If deep learning models are implemented for garage classification, then detection accuracy improves, but computational complexity and processing time increase

Engineering Contradiction:
Improvegarage parking space classification accuracyVSAvoidsystem computational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs preliminary filtering and preprocessing of sensor data before feeding it to the deep learning model, reducing the computational burden while maintaining high classification accuracy for garage parking spaces

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3517409B1Method for detecting garage parking spots
Publication Date: 2024.01.31 VALEO SCHALTER & SENSOREN GMBH
  • EP3517409B1 patent drawingFigure 1
  • EP3517409B1 patent drawingFigure 2
  • EP3517409B1 patent drawingFigure 3

AI summary

In a method for detecting garage parking spaces for vehicles (1) in the vicinity of a vehicle (1) equipped with a parking assistance system (2), wherein the vehicle (1) has at least one environmental sensor (8, 10, 11, 12, 13, 14, 15), the garage parking spaces in a garage (20) are to be reliably detected. This is achieved by providing a method comprising the steps of receiving sensor data with the parking assistance system (2) from the at least one environmental sensor (8, 10, 11, 12, 13, 14, 15) in the vicinity, transmitting the sensor data to a vehicle-side computer unit (6), creating a digital map of the vicinity from the sensor data, recognizing a parking space-like sub-area (21) of the vicinity in the map, and classifying the parking space-like sub-area (21) as a garage parking space using deep learning models.The invention also relates to a parking assistance system (2) for a vehicle (1) to assist a driver of a motor vehicle when parking in a parking space, in particular in a garage (20), and to a vehicle (1) with such a parking assistance system (2).