Error detection for painting robots

With the help of log data, two concrete error patterns are modeled and a procedure for identifying the errors is developed.

Two defect images can be identified with very high accuracy
The software manufacturer has the appropriate instructions for the implementation
Successful presentation of the error detection component to the manufacturer


A software house plans to expand the monitoring software of painting robots of a machine manufacturer to include a component for the early detection of errors. To prove the functionality of early detection in a proof of concept, the software vendor needs working detection models.


A use case workshop is conducted together with the software vendor’s data science team. In this process, significant variables for the error patterns are developed on the basis of the log data (feature engineering). Afterwards, classification models for the detection of error patterns and evaluation of the methods are estimated. On this basis, the steps required for the implementation are described.


The customer can demonstrate the performance of the component to his machine manufacturer using the proof of concept. The software developers have concrete instructions for implementing the error detection.

Are you interested in your own use cases?

Data Science Use Cases

Data & AI Knowledge

Creating joint value from Data & AI


Discover professional articles about Data & AI and current industry news.


Dive into our Best Practices and Industry Exchanges. Discover new dates and recordings of past webinars.


Learn more about the use of Data & AI in your industry through our white papers, case studies and research.