I’m evaluating use of UiPath for automating a Citrix desktop app. But i’m unsure I understand the solution. Should I use OCR-like techniques for Citrix? Or should I be able to indicate specific UI elements like local apps?
Situation:
I’ve installed both UIRemoteRuntime on Citrix en the plugin locally. The two seems to communicate because at first I received a message about mismatching versions. Which I resolved resulting this message is gone. But when I try to indicate an element with the UIExplorer the whole citrix app is shown as one image. Besides this, using OCR technique as shown in Image and Text Automation gives me error messages like Click ‘D3DRenderer2’
Message: Error HRESULT E_FAIL has been returned from a call to a COM component.
@lakshman, thank you for the reply, but in the official documentation I linked to nothing is mentioned about using Computer Vision in relation to Citrix. Is the documentation behind reality in your opinion? And why isn’t it working in the official manner?
Hi, if you want to use computer vision activities
Please install the Uipath.Ai.vision activities package
Then
Use CV screen scope activity
For CV screen scope activity you need a API key which you will find under Platform.uipath.com
Licenses section
For Citrix applications, you generally should not have to rely only on OCR if UiPath Remote Runtime is correctly installed and communicating with the Citrix environment.
Since UI Explorer is showing the entire Citrix application as one image, I would first check the following:
Verify that the UiPath Remote Runtime version matches the UiPath Studio/UiPath.UIAutomation.Activities version.
Restart the Citrix session after installing/updating Remote Runtime.
Confirm that the UiPath Citrix extension/Remote Runtime is properly installed on the Citrix machine.
Check whether the Citrix application is using a supported rendering mode. The D3DRenderer2 / HRESULT E_FAIL error can indicate an issue with the way the application is being rendered.
Try the Computer Vision approach if Remote Runtime cannot expose the individual UI elements.
OCR can be used as a fallback for reading text or locating text-based controls, but it may be less reliable than native selectors/Remote Runtime.
So, if Remote Runtime is working correctly, you should ideally be able to indicate individual Citrix controls, rather than treating the entire Citrix window as one image.
I would first focus on resolving the Remote Runtime / Citrix rendering issue before building the automation entirely around OCR.