Driver Assistance Object Detection via Sensor Fusion

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing driver assistance systems face challenges in accurately identifying objects in close proximity to a vehicle, as image-based object identification methods are limited in near-range scenarios, often failing to detect objects within 0.5 meters, and ultrasonic sensors have narrow detection ranges, limiting their effectiveness.

Innovation Solution

The method combines camera image data with sensor data, using detection algorithms to generate bounding boxes based on both sources, merging them for precise object identification, with sensor data providing distance information and image data offering width, thereby enhancing accuracy and reliability across varying distances.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If object identification is performed solely based on image data, then the system can detect objects at various distances, but the identification accuracy fails in near-range scenarios (up to 0.5m)

Engineering Contradiction:
Improveobject identification accuracyVSAvoiddetection reliability in near range
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent merges image data from a camera with sensor data from ultrasonic sensors to create a hybrid detection system. The computing device processes both data types simultaneously, using sensor data to generate bounding boxes for objects in the near range (0-0.5m) where image-based identification fails, while using image data for objects at farther distances, thereby achieving reliable detection across all ranges.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent introduces sensor data as an intermediary to bridge the detection gap in the near range. When the object identification module cannot identify an object in the image (particularly in the near range), the system uses sensor data from ultrasonic sensors as a mediator to generate bounding boxes and achieve object identification, ensuring continuous detection coverage.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If ultrasonic sensors are used for near-range detection, then detection reliability improves in close proximity, but the detection range remains narrow

Engineering Contradiction:
Improvenear-range detection reliabilityVSAvoiddetection range
Core Design Contradiction:
ReliabilityVSArea of stationary object

Solution Approach 1:

The patent combines the narrow but reliable ultrasonic sensor detection range with the wide but less accurate camera imaging range. The sensor system covers the near range (0-0.5m) with high reliability, while the camera covers the far range (beyond 0.5m). By merging these two systems, the overall detection range is extended while maintaining reliability in both near and far ranges.

Inventive Principle:
Principle #5Merging (Combining)

3Loss of information

If sensor fusion is performed by processing all sensor data in a common computing device, then data integration is achieved, but processing complexity increases

Engineering Contradiction:
Improveinformation integrationVSAvoidcomputing system complexity
Core Design Contradiction:
Loss of informationVSDevice complexity

Solution Approach 1:

The patent segments the object identification process into two distinct modules: an image-based object identification module that processes camera data, and a sensor-based bounding box generation module that processes ultrasonic sensor data. This segmentation allows each module to specialize in its optimal detection range, reducing overall processing complexity while maintaining comprehensive detection capability.

Inventive Principle:
Principle #1Segmentation

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 object identification accuracy and reliability by leveraging the strengths of both image and sensor data, enabling precise detection of objects in close proximity and beyond the ultrasonic range, enhancing collision warning systems and other functionalities.

Implementation Method 1

the at least one sensor is a sensor which operates according to the echo propagation time method

Methodology Applied
Scientific EffectEcho propagation time method: Echo

Data Source

PatentEP2807642B1Method for operating a driver assistance device of a motor vehicle, driver assistance device and motor vehicle
Publication Date: 2019.06.19 CONNAUGHT ELECTRONICS
  • EP2807642B1 patent drawingFigure 1~2
  • EP2807642B1 patent drawingFigure 3~4
  • EP2807642B1 patent drawingFigure 5~6

AI summary

The invention relates to a method for operating a driver assistance device (2) in a motor vehicle (1), in which an image (15) of an environmental region (12) of the motor vehicle (1) is captured by means of a camera (11) of the driver assistance device (2) and in addition sensor data is captured to the environmental region (12) by means of a sensor (3 to 6) different from the camera (11), wherein an object (16) located in the environmental region (12) is identified in the image (15) by means of an electronic computing device of the driver assistance device (2) and the sensor data of the sensor (3 to 6) is used for identifying the object (16) in the image (15).