Artificial Intelligence and Computer Vision applied to Test Automation.
A leading company in the naval sector, operating in 4 continents and 20 countries that offers design and construction services for cruise ships, naval vessels, ferries and mega-yachts, needed to improve the performance of a proprietary CAD desktop application for the creation of 3D models of the products it designs and markets.
The application, used daily by hundreds of designers around the world, caused access and performance problems when faced with high loads, making it impossible to use the tool. The customer therefore needed to be able to simulate, on demand, different ad hoc scenarios to help it understand if its architecture was dimensioned correctly and to collect information on how to tune the system in order to improve its performance under high loads.
For security reasons and in order to comply with corporate policy, the proprietary application involved was only accessible in “virtual” mode via remote desktop or via virtualisation using software such as Citrix. Moreover, it was not possible to directly access elements such as buttons, icons, forms, as would be the case for a normal web or desktop application, with a series of frames/images displayed to the user.
The need was therefore to simulate the use of the application by 100 concurrent users, monitor the execution of ad hoc simulated scenarios in real time and collect useful information in order to carry out a detailed analysis at the end of the simulation.
In order to be able to perform the tests quickly, as well as to verify the different product features, such as the release, it was necessary to automate the test scenarios. We also needed to be able to modify the test scenarios independently, so that the customer would be able to manage them easily.
Thanks to the implementation of Concept Reply's Test Automation solution, it was possible to automate the Performance scenario and to overcome all the existing challenges brilliantly.
Utilising the Visual AI module integrated into the Test Automation Framework, we were able to automate the execution of functional tests on the frames/images delivered by the CAD application.
Through computer vision and deep learning/OCR algorithms, the images generated were subsequently analysed in real time and buttons and icons were identified to interact with the application and to automate the test steps.
The integration of the Test Automation framework was further exploited with docker/container technologies, in order to generate, on demand, up to 100 concurrent virtual instances with which to perform functional tests on the customer’s application.
During the load test, virtual instances were gradually created until the desired values were reached, while unused virtual machines were gradually released to ensure efficient resource management. The test scenarios were designed to be easily extended and modified, as they can be edited in “natural” language, exploiting behaviour-driven programming paradigms.
Concept Reply developed a service that made it possible to automate the execution of Performance Test scenarios on the customer’s virtualised application. The execution of multiple test cycles enabled the replication of the problems encountered by the customer and therefore ensured their correct resolution.
The service will be used by the customer to validate the subsequent releases of their application, together with any architectural revisions of the infrastructure, thus enabling them to guarantee the highest possible quality for their customers.
For today’s companies, Test Automation is no longer an option, it's a necessity. Reply’s new Centre is focused on the Automated Monitoring and Testing of the quality of products and services. Concept Reply has designed an effective Test Automation solution that is scalable and flexible...
Concept Reply is the Reply Group’s hardware and software development partner specialised in IoT innovation. Concept Reply offers solutions in the Smart Infrastructure, Industrial IoT and Connected Vehicle sectors, from the initial idea, through the conceptual phase to implementation, the operational phase and the support phase. The Company’s numerous IoT specialists cover all aspects: from hardware design and development, to software implementation in embedded environments, to edge computing software or Cloud-based applications.