mesh_navigation
mesh_navigation copied to clipboard
naturerobots
Dynamic Obstacle Avoidance
Can mesh navigation plan around dynamic obstacles? Is there some sort of local planner? Or should one from move_base_flex be used?
The dynamic obstacle avoidance is coming in the beginning of next year through a new layer plugin which adjusting the vector field around dynamic obstacles.
Great! You also mentioned on the move_base_flex repo there will be a MBF implementation with grid_maps next year, will it be similar to this? Will this approach still be superior than grid_maps because of these benefits (mainly muli-level environment support), from your paper?
Unlike 2D grid map approaches, this 3D planner works on triangle meshes instead of 2D grid map planes. Working with such 3D surfaces in navigation has the advantage of conformity of the navigation costs based on distances, height differences, roughness, forbidden areas and other criteria. Additionally,meshes represent 3D space directly and can easily be annotated with semantic labels as described in Deeken et al.(2014). Compared to 2.5D solutions, full 3D mesh surface navigation algorithms enable navigating through a multilevel environment, such as multistory buildings or tunnel systems. In particular mesh navigation is appropriate for climbing robots or other robots with the ability to access walls or ceilings.
Hi @jorgemia, It is, since it supports multilayer and multilevel environments. But the efficient and robust mapping of meshes is still and open research topic. The grid_map_nav is planned for next year, yes.
Hello @spuetz,
Hope all is well. I wonder if there are any updates to the new layer plugin which adjusting the vector field around dynamic obstacles?
Looking forward to it!
Hi @spuetz!
I'd like to ask if there are any updates on the dynamic (local) mesh planning?
