ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | Q&A
Ask Your Question

3D Camera Selection

asked 2016-04-19 09:13:06 -0600

Jackie_16 gravatar image

updated 2016-04-19 10:15:36 -0600

I am working on a robot which will pick items from Point A and deliver it to Point B. For this purpose I planned to use a 3D camera and plan the strategy as follows.

  1. Selection of 3D camera [Stereo, TOF or Structured light, Please suggest which one will be the suitable solution?]

  2. Generation of Dense 3D point cloud data, [ By fusion of Laser scanner data and the camera data]

  3. Filter and feature extraction from data , Obstacle avoidance , path planning from Point A to Point B.

If there is any tutorial available which matches with my goals it would be great. Which steps are missing in my strategy ?

Thankyou in advance

edit retag flag offensive close merge delete


I'd recommend editing your question to provide a more descriptive title. People aren't likely to click on a question with a generic title.

jarvisschultz gravatar image jarvisschultz  ( 2016-04-19 10:09:53 -0600 )edit

ok thanks :)

Jackie_16 gravatar image Jackie_16  ( 2016-04-19 10:16:09 -0600 )edit

How much money do you want to spend?

NEngelhard gravatar image NEngelhard  ( 2016-04-19 10:31:55 -0600 )edit

@NEngelhard Lets say for camera part, 300-400$ ?

Jackie_16 gravatar image Jackie_16  ( 2016-04-19 11:19:12 -0600 )edit

1 Answer

Sort by » oldest newest most voted

answered 2016-04-19 10:18:08 -0600

The navigation stack (navstack) would be a good place to start. The navstack provides functionality for path planning, obstacle avoidance, localization, mapping, etc. The navstack is designed to run with a variety of sensors. People have used depth cameras (such as a Kinect) and laser scanners of all varieties successfully with the navstack.

If you are looking to create your own solution, then you are definitely on the right track. Be aware that your step 3 will take quite a lot of effort to get everything working well on a real robot. My advice would be to start with an already-available solution (e.g. the navstack), and then figure out the parts of that solution that you aren't happy with or the parts that you're interested in concentrating your efforts on. Then write your implementations to be compatible with the solution you started with.

The navstack has many great tutorials to get you started.

edit flag offensive delete link more


Great! I understand. Thank you so much.

Jackie_16 gravatar image Jackie_16  ( 2016-04-19 13:06:17 -0600 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools

1 follower


Asked: 2016-04-19 09:13:06 -0600

Seen: 288 times

Last updated: Apr 19 '16