Hi everyone,

I'm currently working on implementing move_base on a quadruped that has the ability to climb stairs. I had managed to configure it for autonomous navigation, meaning I could get it to move to a specified pose given a pre mapped environment. The sensor used is a 2d lidar. I used hectorSLAM for Mapping, AMCL for localization.

I'm now working on adapting my current move_base configuration so that the robot can climb stairs and navigate in another level. I'm currently planning to use different premapped 2D maps to represent different levels, and perhaps a flattened 2D map to represent the staircase path. Perhaps something like that:

Image of Maps: (sorry I don't have enough points) https://drive.google.com/file/d/1B5DnbNt1BUXd-_D1SOnDD1XBlVUjeKyP/view?usp=drivesdk

Is there a way to configure move_base such that the staircase won't be viewed as an obstacle by costmap2d? Thanks!

Note: 1) I've an additional RGB-D camera mounted on the quadruped. I'm not sure if that will help with this use case.

2) I've also got some inspiration from this paper. But it uses a vision based approach to stairs Navigation, though I would prefer a simpler approach. (https://www.sciencedirect.com/science/article/pii/S2405896318332889)