Welding Position Detection Using Heatmap Confidence Correction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional welding robots face challenges in accurately detecting welding positions due to variations in workpiece images, camera performance, surface conditions, environmental factors, and human errors, leading to inaccurate detection and increased man-hours for adjustments.

Innovation Solution

A work system and welding system that utilize an image processing unit to generate heatmap images showing similarity and likelihood distributions, calculating correction amounts and confidence levels for precise welding position detection, with a display unit for confirmation and program correction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If conventional image processing with single method is used, then processing speed is maintained, but detection accuracy and stability deteriorate due to variations in workpiece images, camera performance, and environmental factors

Engineering Contradiction:
Improvedetection accuracyVSAvoidimage processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent combines multiple image processing methods (template matching, feature quantity matching, and deep learning-based detection) into a unified system. Each method generates its own heatmap, and these heatmaps are integrated through synthesis processing to produce a final detection result. This merging approach leverages the strengths of each individual method while compensating for their respective weaknesses, thereby improving detection accuracy and stability without requiring a complete overhaul of the processing system.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The patent creates a composite detection approach by integrating results from multiple image processing methodologies. Instead of relying on a single detection method, the system combines heatmaps generated from template matching, feature quantity comparison, and deep learning models. This composite approach is analogous to using composite materials in engineering, where combining different materials creates a system with superior properties that overcome the limitations of individual components.

Inventive Principle:
Principle #40Composite materials

2Reliability

If multiple image processing methods are combined, then detection stability improves, but processing time and computational load increase

Engineering Contradiction:
Improvedetection stabilityVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent implements a configurable heatmap synthesis approach where the system can selectively apply different levels of processing based on requirements. The synthesis unit can weight and combine heatmaps from various processing methods, allowing operators to adjust the degree of processing applied. This enables the system to achieve sufficient detection stability without always requiring the full computational overhead of all processing methods, thus balancing reliability with processing efficiency.

Inventive Principle:
Principle #16Partial or excessive action

3Measurement precision

If manual adjustment and investigation are performed, then detection accuracy can be improved, but man-hours and operational complexity increase significantly

Engineering Contradiction:
Improvedetection accuracyVSAvoidman-hours
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent implements a self-adjusting detection system where the heatmap synthesis unit automatically combines and weights results from multiple image processing methods without requiring manual intervention. The system self-calibrates by integrating various detection approaches and automatically determining the optimal combination for accurate welding position detection. This eliminates the need for operators to perform manual adjustments and investigations, thereby maintaining high detection accuracy while significantly reducing man-hours and operational complexity.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent incorporates confidence level calculation as a feedback mechanism. The system evaluates the reliability of detection results by analyzing the synthesized heatmap and generates confidence levels that indicate detection quality. This feedback allows the system to automatically adjust processing parameters and verify detection accuracy without manual intervention, thereby maintaining high precision while reducing operational time and complexity.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4653117A1Work system and welding system
Publication Date: 2025.11.26 AMADA CO LTD
  • EP4653117A1 patent drawingFigure 1~2
  • EP4653117A1 patent drawingFigure 3
  • EP4653117A1 patent drawingFigure 4

AI summary

A welding system includes a welding device including a welding head, and configured to move the welding head to a welding position and perform a welding process on a workpiece using the welding head based on an operation program, a control unit configured to execute the operation program to control the welding device, an imaging device configured to image an image of the welding position and its surroundings to output the welding position image, an image processing unit configured to input the welding position image and a reference image at a reference welding position to which the welding head is to be moved according to the operation program, calculate a correction amount between the welding position shown in the welding position image and the reference welding position, and a confidence level of the welding position, and output an image for confirming the welding position, a display unit configured to display the output image, and a program correction unit configured to correct the operation program based on the output correction amount and confidence level.