ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
0

Pre-determined map with RTAB-MAP?

asked 2023-01-08 23:58:44 -0500

chased11 gravatar image

updated 2023-01-09 00:03:14 -0500

I have my robot setup with RTAB-map and robot_localization. The robot is used mainly outdoors (similar to clearpath husky's waypoint_nav), and will pretty much always have a pre-determined map, such as sidewalks and parking lots that I can turn blueprints into a map server. I am trying to determine the better route to take.

Should I trash the idea of using slam since these maps won't really have walls to localize off of, and just use my obstacles from the obstacle_detection package to update my local costmap since my robot will be localized well?

The ekf is tuned well with data from ~3cm GPS, imu, and wheel odom, so localization really shouldn't be a problem. I have had solid results from using the /obstacles for marking and /ground for clearing in a local costmap w/o RTAB-MAP. I have tested both and am leaning towards taking out the SLAM aspect for now. Looking at waypoint_nav, they use hector SLAM for mapping but I am wondering if it's worth it in this aspect.

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
0

answered 2023-01-16 08:49:46 -0500

adityatandon gravatar image

I think you've answered your own question pretty well. If you're confident wrt your localization ability through other methods without SLAM (~3cm in an outdoor environment sounds good to me) and are using a pre-determined map, a simple obstacle avoidance method without the additional overhead of a full SLAM would do.

Also, the SLAM algorithm depends on feature point matching of different features. When you dont have any (or many) feature points, the SLAM process isn't going to be optimal. You could definitely use SLAM, but in this case when you've already solved problems with a simpler approach, is the overhead of a SLAM worth it?

edit flag offensive delete link more

Comments

Thanks for the comment. I ended up using RTAB for mapping and wheel odom+imu in EKF for localization. I was running dual EKF nodes with the GPS for map->base_link but was getting crazy drift, I assume possibly because bad IMU heading data used in navsat_transform, so the accurate gps data was of no use lol. I am still not directly using odometry from RTABmap but it is taking care of the map->odom transform.

This is for a university project and it needs to have some sort of functional autonomy by the end of this week and this is what I came up with for the time being and worrying about fixing the above problems after the presentation.

chased11 gravatar image chased11  ( 2023-01-16 21:04:00 -0500 )edit

Question Tools

1 follower

Stats

Asked: 2023-01-08 23:58:44 -0500

Seen: 87 times

Last updated: Jan 16 '23