Why is there a time delay between the map update and re-planning?

asked Jun 11 '15

sonali gravatar image

Hi everyone,

I'm using the navigation stack with the SBPL Lattice Planner (ARA planner) as the global planner and Trajectory planner as the local planner on a robot. The local cost map can see till 2.5 m in front of it.

When I start the robot and give it a goal a few meters away, it plans a path to it. As it moves, and an obstacle comes within 2.5 m of it, it updates the local cost map and plans a new path. However, there is a significant time delay between the map getting updated and the generation of a new plan because of which the robot sometimes follows its old path till its too near the obstacle to be able to maneuver around it. When I check the SBPL Planner stats that are published, it shows that the re-planning took only 2-3 seconds (the time limit is 3 seconds at the end of which it gives a path even if it is sub-optimal). But time delay between map update and new plan generation is around 4-5 seconds. What is it between the map update and the new plan generation that is taking so much time?

The delay is less when the obstacle is small/narrow and more when the obstacle is wide. So it seems like the part that is causing the delay is dependent on the size of the obstacle.

Please let me know if you need any more information to answer the question.

Preview: (hide)