About the max_waypoint_merge_distance (#340)

[Archive-o-matic]

Posted by @[Missing data]:

Hi @mxgrey

In rmf_demos_fleet_adapter, the max_waypoint_merge_distance is set to 0.5m.
max_lane_merge_distance is 0.1m
If I enlarge the max_waypoint_merge_distance to 1.0m, what is the impact on the traffic planning? Will it be possible to cause collsion or deadlock?
If I enlarge the max_lane_merge_distance to 1.6m. what is the impact?

Thanks
Stella

[Archive-o-matic]

Posted by @mxgrey:

A larger max_waypoint_merge_distance just means the fleet adapter is willing to believe that the lanes of the graph are wider. Consider these pictures where A, B, and C are waypoints and the black lines are the lanes between them:

In the above picture, the lane merge distance is 1m, which only puts it in the range of the lane between A and C. Therefore the robot can start traveling towards A or C, but cannot travel towards B.

If we change the merge distance to 5m, then the robot can immediately travel towards any one of A, B, or C. This might be a problem if there is actually a wall or some other environmental obstacles between the robot and waypoint B. The larger the lane merge distance is, the more likely that the robot will have to deal with problematic obstacles that RMF doesn’t know about. Therefore it’s a good idea to not use an excessively large lane merge distance.

[Archive-o-matic]

Posted by @[Missing data]:

I think you are explaining max_lane_merge_distance.

Then how about max_waypoint_merge_distance? How does RMF use this parameter?

[Archive-o-matic]

Posted by @mxgrey:

The fleet adapter will approximate that the robot is exactly on a waypoint if it is within max_waypoint_merge_distance of the waypoint. I don’t recommend making this value very large because it will create misleading/inaccurate predictions for the planner.