Usage of future obstacle occupancies in the motion planner

CommonRoad-Search
1

Hi,
the motion planner gets the entire scenario including all its obstacles and all the obstacles’ “predictions”. I am not entirely sure whether this “prediction”-object of a dynamic obstacle is meant to be what I (being the steered car in the scenario at some time step t) am assuming where the obstacle will go in the future or whether this is where the obstacle will definitely be in the future.
In the first case, it seems pretty clear that I am allowed to use this data (all obstacles’ predictions in the scenario) in the motion planner’s evaluation_function(), evaluating the state of the car in the next time step, right?
In the second case, it seems like I (more specifically the motion planner at some time step t) am not supposed to know these future movements and should not use them in the motion planner, right?
So my question is: Am I allowed to use the prediction-data in the motion planner’s evaluation_function() at every time step t?

Thanks!

2

Hi Lauer,
Yes, you are allowed to use this information. The provided prediction object is the actual trajectory recorded from its source, so the obstacle with definitely move according to the prediction.
In real application, this information is unknown to the motion planner, and thus it is usually accompanied by a predictor to predict the future motion of other objects in the scenario.

1 Like