ML-Based Constraint Generation for CAD Assemblies

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Determining and specifying constraints in modern CAD contexts is time-consuming, resource-consuming, and prone to errors, especially when dealing with complex assemblies comprising hundreds of thousands of CAD parts.

Innovation Solution

The implementation of machine-learning (ML) based systems for the generation of constraints in CAD assemblies, which includes a constraint learning engine that generates a representation graph from CAD assemblies and a constraint generation engine that applies an ML model to predict and generate constraints without the need for explicit placement of CAD parts.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual methods are used to determine and specify constraints in CAD assemblies, then users have full control over constraint specification, but the process becomes time-consuming and resource-consuming

Engineering Contradiction:
Improveconstraint specification accuracyVSAvoidconstraint determination time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system enables self-service by allowing the CAD assembly data to automatically generate constraints through machine learning models. The constraint generation engine processes geometric relationships and automatically determines appropriate constraints without requiring manual user intervention, thus resolving the contradiction between reliability and time consumption.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces the mechanical manual process of constraint specification with an automated computational system. Machine learning models analyze geometric relationships and automatically generate constraints, substituting the manual mechanical process with an intelligent automated system that maintains accuracy while dramatically reducing time requirements.

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

2Reliability

If manual methods are used to specify constraints in complex assemblies, then users can ensure proper constraint application, but the process becomes prone to errors

Engineering Contradiction:
Improveconstraint application correctnessVSAvoidconstraint specification complexity
Core Design Contradiction:
ReliabilityVSEase of operation

Solution Approach 1:

The system performs self-service by automatically analyzing geometric relationships and determining appropriate constraints without human intervention. This eliminates human error while maintaining correctness, as the machine learning models are trained to accurately identify and apply appropriate constraints based on geometric features.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system incorporates feedback mechanisms where the machine learning models continuously learn from training data consisting of correctly constrained CAD assemblies. This feedback loop improves the accuracy and reliability of constraint generation over time, ensuring proper constraint application while reducing operational complexity.

Inventive Principle:
Principle #23Feedback

3Productivity

If automatic constraint generation is implemented using machine learning, then the process is accelerated and simplified, but the initial setup and training requirements increase device complexity

Engineering Contradiction:
Improveconstraint generation speedVSAvoidsystem architecture complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The system applies preliminary action by pre-training machine learning models on extensive datasets of CAD assemblies with known constraints. This preparatory training phase, while complex, is performed once beforehand, enabling rapid and simple constraint generation during actual CAD operations, thus achieving high productivity while managing device complexity through advance preparation.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent introduces an intermediary machine learning model that sits between the raw CAD geometry and the constraint specification process. This intermediary component handles the complexity of constraint determination internally, presenting a simplified interface to users while maintaining high productivity through automated processing.

Inventive Principle:
Principle #24Intermediary (Mediator)

4Ease of operation

If machine learning models are used to predict constraints, then user burden is reduced and efficiency improves, but computational resources and processing time increase

Engineering Contradiction:
Improveuser effort in constraint specificationVSAvoidcomputational energy consumption
Core Design Contradiction:
Ease of operationVSUse of energy by moving object

Solution Approach 1:

The system applies partial action by focusing the machine learning model's computational effort only on the specific geometric features relevant to constraint determination, rather than analyzing the entire CAD assembly in full detail. This selective processing reduces computational energy consumption while maintaining ease of operation and efficiency.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP4185981B1Machine learning-based generation of constraints for computer-aided design (CAD) assemblies
Publication Date: 2025.04.23 SIEMENS INDUSTRY SOFTWARE INC
  • EP4185981B1 patent drawingFigure 1
  • EP4185981B1 patent drawingFigure 2
  • EP4185981B1 patent drawingFigure 3

AI summary

A computing system (100) may include a constraint learning engine (110) and a constraint generation engine (112). The constraint learning engine (110) may be configured to access a computer- aided design (CAD) assembly (130) comprising multiple CAD parts and generate a representation graph of the CAD assembly (130), determine constraints in the CAD assembly (130), wherein the constraints limit a degree of movement between geometric faces of different CAD parts in the CAD assembly (130), insert constraint edges into the representation graph that represent the determined constraints; and provide the representation graph as training data to train a machine-learning model (120). The constraint generation engine (112) may be configured to generate constraints for a different CAD assembly by applying the machine-learning model (120) for the different CAD assembly.