I’m trying out the navigation tutorial here: Nav2 Basics - Stretch Docs (hello-robot.com). I was able to create a map, but when I set a 2D Nav Goal, the robot fails to create a path to the goal and reach it, even though the goal is very close to the robot’s initial location and is reachable with a near-straight line path. I think the issue might be that there are several false obstacles created by the mapping algorithm that aren’t actually present (the robot is in a clear space and yet there are black dots and cyan areas on the map where there shouldn’t be). Any ideas on how to fix this?
Hi @ycode welcome to the forum, looking at the issues you have described here are my comments:
- Regarding the black dots/false obstacles present in the map
- If you inspect only the Lidar data visualization live do you see false positive obstacle ?
- While mapping could you make sure the lidar is not blocked by any other objects attached to the robot?, usually we have seen either robot’s gripper itself blocking or any large items attached to the robot’s trunk USB hub blocking the lidar which gives out false positives.
- Try Adding a Denoise Layer to the Global Cost Map
- You can try adding denoise layer to the global costmap using this documentation on filtering noise-induced obstacles Nav2. In stretch_ros2 you will need to be adding your Denoise layer to the file nav2_params.yaml