Intelligent test automation – in the automotive sector
Development of a test planning software using Natural Language Processing & Supervised Machine Learning.
Follow us
The future
of automotive engineering
A renowned German car manufacturer is a leading manufacturer of premium automobiles. The company operates over 30 production sites worldwide and is pursuing an ambitious electrification strategy with the aim of achieving a 50% sales share of fully electric vehicles by 2030.
The company has strict quality assurance requirements for the software used in its cars. Complex tests that cover all of a vehicle’s 40 or so control units require a great deal of manual effort, as they were still laboriously selected and planned manually by large teams until the new software was rolled out. The aim was to optimize this time-consuming and cost-intensive test planning by an intelligent software solution.
.
The Challenge:
Automate the test planning and reduce the test effort.
Testing is carried out at every stage of the software development and process level to ensure that the components function properly. However, frequent multiple tests are inefficient and should be improved through test automation, so that testing is only repeated in the event of innovations or changes. In addition, many different departments are involved in the construction of a vehicle, all of which use different test descriptions, resulting in a large number of test cases. Therefore, higher test levels often do not know whether a function has already been tested at a lower level.
The solution:
The rollout of an efficient software thanks to intelligent test automation from QualityMinds
QualityMinds develops innovative test planning with learning algorithms and automated determination of test coverage.
In a discovery workshop, QualityMinds supported the customer in understanding the problem, defining it more precisely and describing it in detail. User stories and usability studies were used to determine the exact needs of the users (in this case, the car manufacturer’s testers) for the desired software. Within just four months, QualityMinds succeeded in creating a first version of the product, which was continuously optimized through further usability studies and feedback loops and was tailored to the daily work of the testers.
Key Facts
Automated selection of test cases reduces test effort and optimizes the test planning
Technologies: Java, Machine Learning, MariaDB, Docker, REST, d3 Library, Bootstrap, Spring Boot, integration of the test environment in Jenkins
Compliance with security requirements has top priority: development at QualityMinds but deployment locally at the customer’s site via their intranet
Expertise in agile software development, design thinking and testing from a single source at QualityMinds
Use of the Vector Space Embedding technique and a co-occurrence matrix in the context of NLP:
- Vector Space Embedding: Conversion of words into vectors to analyze semantic similarities in test cases
- Co-occurence Matrix: Calculation of the frequency of common words in the analysis of test cases
“We developed the software with the help of Kanban tools in an agile manner, continuously gathering feedback from the testers as to whether what we were developing was actually helping them in their daily technical work. When the software went live, it made the testers’ daily work much easier. One of QualityMinds’ strengths is that, thanks to discovery workshops and extensive usability studies during the project, we can precisely understand the needs of the users, quickly implement the client’s change requests and thus develop first-class software.”
Tobias Varlemann
Project Manager Head of IT at QualityMinds