Automating VDI- AI Computer Vision put into Action
December 19, 2019
Virtual Desktop Infrastructure (VDI) – once a far-fetched dream for technology enthusiasts has now become an imminent reality. The server-based computing technology that hosts a desktop operating system on a centralized server in a data center has now become the corporate favorite, for its hardware, licensing and capital savings, storage consolidation, data security to name a few. It is a sturdy case of virtualization as it provides personalized desktops for each user with all the boons of centralized management. The big guns in the business include Citrix, Microsoft and VMWare.
KGiSL, keeping up with the latest technology, has been always inclined towards inducing newest trends in their process and our latest interests lie in practicing the scope of RPA in VDI automation. Our success story in insurance sector indicates the wide scope of RPA implementation in VDI. KGiSL automated data entry process of an insurance vendor using the computer vision feature of UiPath.
Here is the demo video of the automated process.
We faced our share of setbacks in the initial stages and one of the major challenges included the efficiency in automating VDI environment. UiPath bots can identify the application elements (such as buttons, text-fields, etc.) using their unique selectors when automating through native desktop systems. Bots cannot identify individual elements of the Applications which are accessed through the VDI and treats the entire VDI as an image. The developer is left with no choice but to automate using image comparison techniques, where bot searches for images of application elements to perform actions such as clicks, types into, etc. This technique is not accurate at all the times, leaving the entire process less reliable and unfit for fool proof automation. Many recent advancements using OCR technology, image-based automation, and coordinates-based automation have been proved ineffective as these attempts have led to reliability and maintenance issues, since even minor changes in the UI will break the automation.
When there was simply no way out of this conundrum, UiPath technology innovators introduced an infallible method for automating VDI environments using AI, Machine Learning (ML), Natural Language Processing (NLP) and Computer Vision (CV). Dynamic linking between objects has been subtly introduced which determines their composition. This results in providing accurate identification of application elements despite changes to UI. This ensured a human-like recognition of user Interfaces, using a mix of AI, OCR, text fuzzy-matching, and an anchoring system to bring it all together. This method can also be swiftly extended to other cases of SAP, Flash, Silverlight, PDFs, and even images. It is far away from traditional image automation as it does not get affected by any changes in color, font, size, and resolution of the Infrastructure.
Even though modern technologies have increased the efficiency of automating VDI, we are still in the inceptive stages and we have devised productive strategies to keep it moving. Our roadmap has included scope for usability improvements to this current version, support for recording full automation using AI Computer Vision and ultimate level of capability.