Robot Process Optimization Using Learned Quality Assessment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for optimizing robot processes struggle to formulate suitable quality criteria, particularly in complex or difficult-to-evaluate tasks like robot-assisted clipping in of snappers, where success cannot be easily read from force or joint angle data.
Innovation Solution
A method and system that utilize machine learning to assess process run-throughs and optimize robot process controls. This involves executing process run-throughs with varying process controls, detecting assessments and process variable values, and machine-learning a quality factor model to improve process optimization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If traditional optimization methods are used with force angle courses or joint angle courses, then the optimization process can be automated, but the quality criterion cannot be accurately determined for complex processes like robot-assisted clipping in
Solution Approach 1:
A machine learning model is introduced as an intermediary between the robot process and the quality assessment. The model learns to map process parameters (force angle courses, joint angle courses) to quality criteria by training on expert assessments, enabling automated determination of quality for complex processes where direct measurement is difficult
Solution Approach 2:
The traditional direct measurement approach for quality criteria is replaced with a machine learning-based prediction system. Instead of relying on straightforward sensor readings, the system uses trained models to infer quality metrics from process data, substituting mechanical measurement with intelligent computation
2Reliability
If multiple process run-throughs are executed for optimization, then the quality factor model can be trained, but the time and number of run-throughs required increases
Solution Approach 1:
Expert assessments are collected in advance during the training phase to create a training dataset. This preliminary action allows the machine learning model to learn from pre-collected expert knowledge, reducing the need for extensive iterative run-throughs during the actual optimization process
Solution Approach 2:
The system implements feedback loops where expert assessments of process run-throughs are continuously fed back into the machine learning model for training and refinement. This feedback mechanism allows the model to improve its predictions over time, reducing the number of run-throughs needed to achieve reliable optimization results
3Measurement precision
If expert assessments are used for quality criterion determination, then accurate quality evaluation is achieved, but the process cannot be fully automated and requires human involvement
Solution Approach 1:
The machine learning model creates a computational copy of expert assessment capabilities. By training on multiple expert assessments, the model learns to replicate human judgment and evaluation skills, enabling automated quality assessment that mirrors expert-level accuracy without requiring continuous human involvement
Solution Approach 2:
The system enables self-service automation where the machine learning model independently performs quality assessment without requiring real-time expert intervention. The model serves itself by using its trained knowledge to evaluate process run-throughs, freeing experts from routine assessment tasks while maintaining high accuracy
Data Source
AI summary
A method for executing a process, in particular using at least one robot, includes executing a run-through of the process, detecting a value of a first process variable, and detecting an assessment of this executed process run-through. Assessment learning steps are then repeated multiple times, wherein run-throughs of the process using varied process controls are executed and additional assessments are detected. A first quality factor model of the process, which model determines a quality factor for the process on the basis on the first process variable, is machine-learned based on the detected assessments and values of the first process variable. The method further includes repeating process control optimization steps multiple times.