Re:Gobot: A framework for robotics, physical computing, and ...
by Symmetry on Thursday December 25, 2014 @07:53AM
(#8796157)
Right, though I think it would be fairly straightforward to make a bridge. This is all pretty cool, but I think I'd really want something lower level than a RESTful API for streaming data from a SLS or a LIDAR to assemble into a costmap. I see they've included OpenCV in their system but no PCL[1]. I suppose that makes sense since you can just connect a camera to your OpenCV handler directly but when you're using pointcloads you're usually combining a bunch of them.
Of course, when I create my Nim or Rust based robotics framework in my Copious Spare Time I'm going to use a pub-sub network where you can declare topics to be local-reliable (shared memory), global-reliable (tcp), or global-unreliable (udp-based).
Re:Gobot: A framework for robotics, physical computing, and ...
by Qworg on Thursday December 25, 2014 @09:07AM
(#8796251)
I like that vision of pub-sub, BTW. It helps to make things explicit (and bar access when it would conflict).
I agree that PCL integration is a must for larger robot platforms.
I'm unsure I understand their "platforms" argument - OpenCV as platform?
Re:Gobot: A framework for robotics, physical computing, and ...
by dom96 on Thursday December 25, 2014 @10:40AM
(#8796431)
It's nice to hear that you are considering using Nim for this.
Re:Gobot: A framework for robotics, physical computing, and ...
by ricw on Thursday December 25, 2014 @12:34PM
(#8796698)
ROS has an entirely different target group. You'll never be able to run ROS on a traditional robotics / IoT framework. We're talking KBs of RAM; kilo/megahertz CPU speeds; minimal power consumption.
[1] http://www.ros.org/
[2] http://wiki.ros.org/Robots
(Edit: formatting)