Should I use a ROS object viewer for natural language processing research?

asked 2015-07-25 09:16:32 -0600

I am working on natural language understanding and am looking for artificial intelligence components to integrate with. I need an image viewer. I was thinging of using a ROS viewer. As part of language understanding you need to be able to view and refer to objects and have a sense of 3D space. The words this, that, those, and the cannot be understood without the context of an object in a viewer. I am thinking my natural language understanding would be helped by a virtual robot that could potential interact with other virtual robots or even humans.

I want to be able to construct views of objects based on text input and knowldegebase information. Ideally I would like to annotate objects I create and have the annotations on different properties on different viewer layers and ideally move information between layers. That way I can extract text information a decide what is virtual / abstract or a real world object or property in one viewer.

I do not currently use ubantu but would not be adverse to switching to it. I would be happy to use an online service for this with a view of moving to an an embedded / embodied system in the future.

Is this going to be possible ? Is there a better approach ? Have you any tips? My project page can be found at www.hemseye.net .

edit retag flag offensive close merge delete