ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
1

inferring missing objects

asked 2012-10-11 12:46:35 -0500

salma gravatar image

updated 2012-10-14 13:32:18 -0500

Hi, i begin to do asimple robot that can inferring missing objects on the table like that :image description

firstly i want to list all i need to do this 1- kinect : can i use any type of kinect or must openni kinect ? or can i use a normal camera ?? why ?

2- the simulation in the picture up is it gazebo ?? how can i link my simulation with reasoning and knowrob ??

3- the relation between my simulation and the picture captuered by my kinect or camera ?

Thanks :)

edit retag flag offensive close merge delete

Comments

too little info. where do you get this picture?

yangyangcv gravatar image yangyangcv  ( 2012-10-11 13:47:46 -0500 )edit

it is a simple target using kinect or camera to just inferring missing component using semantic mapping for objects ,this picture from pdf in semantic

salma gravatar image salma  ( 2012-10-11 23:45:49 -0500 )edit
salma gravatar image salma  ( 2012-10-12 00:07:49 -0500 )edit

1 Answer

Sort by ยป oldest newest most voted
5

answered 2012-10-13 20:53:51 -0500

moritz gravatar image

Though inferring the missing objects may appear simple, it is in fact quite a complex problem that requires the integration of perception, knowledge representation and statistical relational models. You can have a look at our paper "Combining Perception and Knowledge Processing for Everyday Manipulation" to get an idea of the techniques we have used.

The corresponding ROS packages are comp_missingobj, prolog_perception, mod_probcog, srldb, and the knowrob stack. The experiment was done in early 2010, however, and I am not sure if all components still work in current ROS versions, especially prolog_perception may make problems since we haven't used it in a while. In addition, you will need an object recognition system that can give you the types and positions of objects.

The visualization is not gazebo, by the way, but the visualization module of the knowrob knowledge base (package mod_vis).

edit flag offensive delete link more

Question Tools

2 followers

Stats

Asked: 2012-10-11 12:46:35 -0500

Seen: 279 times

Last updated: Oct 14 '12