Navigation Stack Computation from External Computer

edit

Continuing off the comments, I dont know how your system/packaging is setup, so its hard to say if you want to take the dependency off or not. For example, if you have say 1 github repo that you clone to both the beaglebone and the external computer so that they both have the same files, then you probably wouldnt want to remove move_base as a dependency since it will also affect your external computer.

Ideally, I think you may want to setup your system in a way similar to how Clearpath organizes there github repo for their jackal robot . You would basically divide the different parts related to the robot into separate repos. For example you could have say two repo's:

• robot_control (this repo could contain any driver/sensor related items)
• robot_navigation (this repo would have your move_base related items)

That way you could only clone the robot_control repo to your beaglebone (and not even have to install/have move_base on the beaglebone) and clone the robot_navigation repo to your external computer (where you would have move_base listed as a dependency just like they tell you to in the setting up the navigation stack tutorials)

If you setup your network correctly, then just run the sensors and robot driving related stuff on your beaglebone, and run move_base on your external computer. You can also look at the roslaunch machine examples to launch nodes on different machines, from a single computer (all on same network)

EDIT 1: My understanding of it (may be wrong) is that "jackal" is a common package used amongst all the different types of jackal setups (simulation or real world). Things like the URDF of the robot, and custom messages the robot sends. Jackal_robot is specific to the physical robot itself (boot up/onstart procedures). If you spend some time looking through clearpath's repo's you will learn ALOT. It just takes time and curiosity :)