You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi, I noticed that when you are collecting training data using autopilot, your target point are the far_nodes coming from _command_planner. However, when your actual model based agent is tested, it uses the _waypoint_planner to get target point which corresponds to near node. I think it is better to use the near nodes both for training and testing? Or do I miss something?
The text was updated successfully, but these errors were encountered:
near_node is the next waypoint after 4 metres whereas the far node is the next waypoint after 7 metres. During testing, I don't think it matters much whether you use near node or far node since the waypoints along the route are quite far from each other. But, I agree that it's better to keep it consistent between training and testing.
Hi, I noticed that when you are collecting training data using autopilot, your target point are the far_nodes coming from _command_planner. However, when your actual model based agent is tested, it uses the _waypoint_planner to get target point which corresponds to near node. I think it is better to use the near nodes both for training and testing? Or do I miss something?
The text was updated successfully, but these errors were encountered: