[ros-users] [Discourse.ros.org] [ROS Projects] Autonomous r…

Forside
Vedhæftede filer:
Indlæg som e-mail
+ (text/plain)
Slet denne besked
Besvar denne besked
Skribent: Cees Trouwborst via ros-users
Dato:  
Til: ros-users
Emne: [ros-users] [Discourse.ros.org] [ROS Projects] Autonomous rail-guided robot with manipulator



Some update on my progress! All input is highly appreciated. The video below shows the current state of affairs (turn on Closed Captioning for some explanation): I am able to move the robot over a piece of rail in simulation, while RViz tracks it position. Very happy with it, because for now it does exactly what I need. :slight_smile:

https://www.youtube.com/watch?v=CRyQFCoVTh4

Some remarks, though:

* The ball joints between the different carts are modelled using three continuous joints and two dummy bodies. This is far from elegant, but MoveIt! and Multi-DOF joints resulted in more problems than I could solve. @v4hn and I talked about alternative approaches over IRC, and if time allows, I'll definitely try to fix this.
* Allowing collisions between some parts instead of modeling the carts on the rails was a nice idea, but it didn't work. Apart from the collision mask troubles, it resulted in messy and unpredictable situations because the dynamics would still be active. So, when one body moves, the other bodies want to move as well and do not stay aligned with the rails nicely. I simplified the collision meshes and Gazebo actually simulates the carts sliding over the rails.
* Collision objects are apparently always remapped to the base link of the robot. So... if the robot moves, the collision objects move with it. While writing this, I remember that this might be something that I can fix with a virtual floating link. That is a better solution than just republishing the objects, which is a waste of resources and results in jittery motion. This might also make it easier to switch back to using an Octomap, which means I can use raycasting.
* Performance of the simulation is adequate, I think. It is far from real time, I expect the main reason for this is the number of contact points between the carts and the rail. That seems pretty hard to accurately simulate.






---
[Visit Topic](https://discourse.ros.org/t/autonomous-rail-guided-robot-with-manipulator/966/2) or reply to this email to respond.


If you do not want to receive messages from ros-users please use the unsubscribe link below. If you use the one above, you will stop all of ros-users from receiving updates.
______________________________________________________________________________
ros-users mailing list

http://lists.ros.org/mailman/listinfo/ros-users
Unsubscribe: <http://lists.ros.org/mailman//options/ros-users>