Ultrasonic Object Position Detection Using AI Region Analysis

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for detecting the position of objects using image recognition are inefficient due to high computational requirements and difficulties in separating objects from light effects and backgrounds, making it challenging to determine if objects are in the correct position with minimal computing power.

Innovation Solution

A method utilizing ultrasonic sensors to measure objects with distinctive components, converting ultrasound reflections into electrical signals, and using artificial intelligence to evaluate only specific areas for comparison with reference images, allowing for quick detection of correct object positioning without needing high-powered computers.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If image recognition technology is used to detect object position, then object identification can be achieved, but high computational power and complex processing are required

Engineering Contradiction:
Improveobject position detection accuracyVSAvoidcomputing power requirement
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent divides the object into multiple regions of interest (ROIs) based on distinctive components. Instead of processing the entire object image, only specific regions containing distinctive features are extracted and evaluated. This segmentation approach reduces the computational load while maintaining detection accuracy, as the AI only needs to analyze relevant portions of the object rather than the complete image.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts only the necessary information from the object image by identifying and isolating regions containing distinctive components. By taking out only the relevant regions for evaluation and comparing them with reference images, the system avoids processing unnecessary data, thereby reducing computational requirements while preserving the essential information needed for accurate position detection.

Inventive Principle:
Principle #2Taking out (Extraction)

2Reliability

If image recognition processes evaluate the entire object, then complete object analysis is achieved, but processing time increases and computational power is consumed

Engineering Contradiction:
Improveobject position detection reliabilityVSAvoidmeasurement speed
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent segments the object evaluation process by identifying multiple regions of interest containing distinctive components. Each region is processed independently and in parallel, which maintains comprehensive object analysis reliability while significantly reducing total processing time. This segmented approach allows the system to evaluate critical features without being burdened by unnecessary data from irrelevant object areas.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies partial action by evaluating only the necessary regions of the object that contain distinctive components, rather than processing the entire object. This selective evaluation maintains sufficient reliability for position detection by focusing on critical features, while dramatically improving productivity by avoiding processing of redundant information.

Inventive Principle:
Principle #16Partial or excessive action

3Measurement precision

If image recognition is used to separate object from background and light effects, then accurate object identification is possible, but the process becomes extremely difficult and computationally intensive

Engineering Contradiction:
Improveobject separation accuracyVSAvoidprocessing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent uses segmentation to divide the image processing task into manageable regions of interest. By isolating specific areas containing distinctive components, the system avoids the complexity of separating the entire object from background and light effects. This regional approach simplifies the processing while maintaining accurate object identification, as each ROI can be evaluated independently with reduced computational complexity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts only the essential regions containing distinctive components from the complete image, removing the need to process background and light effect interference across the entire image. This extraction approach maintains object separation accuracy by focusing on clean, distinctive features while significantly reducing processing complexity and computational requirements.

Inventive Principle:
Principle #2Taking out (Extraction)

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

Enables rapid measurement and sorting of objects using minimal computing power, ensuring only correctly positioned objects are processed further, as the evaluation focuses on specific marker areas rather than the entire object, facilitating efficient use of simple AI and automated sorting systems.

Implementation Method 1

the objects to be measured (1, 8) are measured (1.3) by means of ultrasound (1.3), wherein the ultrasound waves are reflected by the objects (1, 8) and are picked up again by the at least one ultrasonic sensor (24 to 27) and converted into electrical signals

Methodology Applied
Scientific EffectUltrasound reflection: Reflection

Implementation Method 2

the ultrasonic sensors (24 to 27) emits the ultrasound waves and pick up the reflected ultrasound waves and convert them into electrical signals

Methodology Applied
Scientific EffectPiezoelectric effect: Piezoelectric Effect

Data Source

PatentEP3396404B1Method for locating objects
Publication Date: 2020.02.19 PEPPERL & FUCHS GMBH
  • EP3396404B1 patent drawingFigure 1
  • EP3396404B1 patent drawingFigure 2
  • EP3396404B1 patent drawingFigure 3

AI summary

A method for object position detection is described, comprising the following successive steps. In a first step, several objects (7 to 10) are provided, each possessing at least one distinctive component (12 to 15). In a second step, a first object (8) is transported to a measuring station (5) in which at least one ultrasonic sensor is located. In a third step, the object (8) is measured using ultrasound, whereby the ultrasonic waves are reflected by the object. These ultrasonic waves are received by the at least one ultrasonic sensor and converted into electrical signals.In a fourth step, the electrical signals originating from the area (17) where the distinctive component is located are sent to an artificial intelligence (23) and then converted into an image. Finally, in a last step, the resulting image is compared with a reference image to check whether the distinctive component (13) is located in the evaluated area (17). If the distinctive component (13) is located in the evaluated area (17), then the object (8) is in the correct position.