Running navigation stack with odom, IMU and gps

Is it possible to run the navigation stack and navigate towards goals (ignoring obstacles) by combining odometry, IMU data and gps data? The robot is meant to drive in an open field where basically everything looks the same and there are barely any obstacles. Beacons or any kind of fixed frame of reference aside from gps isn't available.

Asked by Gherkins on 2021-01-12 22:17:09 UTC

Comments

Answers

It's more of a localization problem, then navigation. The precision of localization depends on quality of your sensors
If you're not afraid of obstacles you don't have to use them in planning or you can route a predefined paths (assumed your localization quality is fine)

Asked by soldierofhell on 2021-01-13 02:31:37 UTC

Comments

As mentioned by @soldierofhell this is more a "localization" problem than a navigation problem. In order to combine odom, IMU, and GPS to generate an estimate of your position consider using the robot_localization package. This, as far as my experience is concerned, is the standard way to do what you want to do. Once you have your robot localized in the space then the ROS nav stack will be able to take you from waypoint to waypoint.

For the integration of GPS robot_localization provide the navsat_transform_node and a detailed tutorial on how to integrate this with the other sensors to generate a localization state estimate.

Asked by JackB on 2021-01-13 08:54:12 UTC