Using PointCloud data for object detection

Question

Hello,

I'm working on a project that uses Kinect as sensor for a robot. I intend to use PointCloud library for ROS. However, I don't know how to resolve or use the PointCloud data in order to detect objects. The object detection will be used in order to avoid obstacles using potential fields principle. So, I need to transform PointCloud data to obtain all possible obstacles (their coordinates, width, height, distance) in order to create forces to be used in the robot navigation. I don't know which tutorial should I follow? I would really appreciate if anybody gives me some guidance. Thanks in advance.

---

Originally posted by mozcelikors on ROS Answers with karma: 181 on 2013-07-29

Post score: 1

Answer 1

I don't know if something like this already exists but I suggest you to use Octomap to have a collisionMap. Maybe with that, you can then isolate obstacles easier. See this link

You will need to have a filter to erase the ground, this octomap server has this too.

EDIT: If you know the object, you can use tabletop_object_detector but I don't believe this is what you are looking for.

---

Originally posted by jep31 with karma: 411 on 2013-07-29

This answer was ACCEPTED on the original site

Post score: 2