Revision history [back]

I was trying to make a simple nav_goal to cmd_vel generator outside of the move_base architecture and in python but still using as much of the supporting components as possible. I ended up bringing up costmap_2d somewhat similar to http://answers.ros.org/question/185605/getting-costmap_2d-to-update-with-laserscan-data and have a python node that subscribes to the nav goal and transforms it to the base_link frame, generates a cmd_vel directly from that, and calculates the cost of the costmap of the overlap of the footprint and the subscribed costmap (which has a low 160x120 resolution and low update rate, so not a lot of performance problems there) and slows down the cmd_vel when cost is increasing and too high.

A while back I was trying to get the full move_base working, and it sort of did (it never commanded linear_y even though I configured it to be holonomic) but I decided it was easier to go with something more crude until I could devote enough effort to move_base. But I might also use OMPL in python to add obstacle avoidance as another intermediate step.