ROSIntegrationVision for Unreal Engine 4

This plugin adds ROS Vision Support to your Unreal Engine 4 Project. A specialized Camera can measure rgb and depth data from your Unreal World and publishes it into a running ROS environment. In order to use this plugin you also need to add the ROSIntegration Core Plugin (https://github.com/code-iai/ROSIntegration) to your project.

Dependencies of this Plugin

This plugin depends on F16C Intrinsic Support (https://msdn.microsoft.com/de-de/library/hh977022.aspx), which should be included in newer CPU generations. This ensures the depth data to be converted quickly. Important: If you use this Plugin on Linux, you need to enable F16C support manually in Unreal Engine 4 and recompile it. To do this, open PATH_TO_UNREAL/Engine/Source/Programs/UnrealBuildTool/Platform/Linux/LinuxToolChain.cs, find the GetCLArguments_Global method and add Result += " -mf16c"; in a suitable place. After that, recompile UE4.

Usage

After installing this plugin and the core ROSIntegration plugin, you can load your UE4 project. If the plugin is not enabled in the project yet, you can do this in the UE4Editor in Edit -> Plugins.

When the plugin is loaded correctly, a new Actor named VisionActor and a new CameraComponent named VisionComponent will be available to use. Each one represents the RGBD camera.

In some cases (for example on Linux), it might be necessary to call the Generate Project Files action on UE4 in order to fetch the new header files for the plugin before your first compile.

Vision Component

Actor Attachment:

#include "ROSIntegrationVision/Public/VisionComponent.h"
...
UVisionComponent * vision = CreateDefaultSubobject<UVisionComponent>(TEXT("Vision"));
vision->SetupAttachment(RootComponent);

Disable TF Publishing:

This is useful for fixed cameras with links / joints defined via URDF.

vision->DisableTFPublishing = true;

Camera Parent Link Naming:

vision->ParentLink = "desired_link"

Vision Actor

A bare-bones Actor with a VisionComponent attached to it's RootComponent

Credits

Credits go to http://unrealcv.org/ and Thiemo Wiedemeyer, who laid out the rendering and data handling basics for this Plugin.