How to trigger navigation after initial pose estimation ?

[soblin]

We are trying to use the following launch file, in which the robot first estimates its initial pose using the base map and the scanned points.

https://github.com/tier4/autoware_launch/blob/tier4/universe/autoware_launch/launch/autoware.launch.xml

We prepared an indoor 3D-map by running SLAM with a mobile robot attached with a velodyne lidar, so we already have access to base map and scan points. Then in the Rviz we give an initial guess of the robot pose manually by clicking "2D Pose Estimate" button on the toolbar. It works fine, and the scanned points and the map look well fitted (see attached image).

In the next step, we want to give a goal to the robot and then start navigation from the initial pose by clicking the "2D Goal Pose" button just like we do with planning_simulator, but nothing happens.

So is there any way we can try out robot navigation with this launch file ?

[TakaHoribe]

The next step should be to publish the goal to the mission_planner. But some concerns exist. I'm not 100% sure though,

• The robot should be inside the lane when you publish a goal
• The goal lanelet (where the goal is located) and the initial lanelet (where the robot is located) should be separated. If they are the same, route planning will not work correctly. In that case, you have to separate the long lanelet into some small lanes.

[soblin]

Thank, actually it was due to another reason, but thanks for your helpful answer!

[g-ise]

Another possibility is that some necessary topics are missing and Autoware's STATE does not change from INITIALIZING.
You may see the state by adding an rviz panel AutowareStatePanel