[Feature request]: Robot stops at every point.

[saksham07sk]

Before proceeding, is there an existing issue or discussion for this?

• I have done a search for similar issues and discussions.

Description

I have added a curve lane feature into traffic editor as there are multiple points and in between them there is straight lines to form a curve, while navigating through that points robot stops at every single point and that must not to be happen.
As per my knowledge slotcar forces robot to stop at point and corrects the direction as per next waypoint.

How can I tackle this issue ?, I want that robot must follow curve path (angular path) as nav2's /follow_through_pose
Following is video of issue.

Stuck_behaviour.-.Made.with.Clipchamp.mp4

Implementation Considerations

No response

Alternatives

No response

Additional information

No response

[mxgrey]

The slotcar plugin isn't particularly designed to handle curved paths. It's only meant to be the simplest possible simulation of a vehicle following commands from rmf_fleet_msgs. It doesn't have any navigation stack built in because the slotcar is meant for multi-agent simulations, and we found early on that most computers cannot handle running an entire nav stack for multiple robots at once without slowing the simulation to a crawl.

The way to get more complex trajectories in Open-RMF is to specify a dock name on the vertex that you are trying to reach and then handle that with some custom logic where you command the robot along the path (e.g. call a /navigate_through_poses action for the robot). There are two caveats with this:

1. A vertex with a dock parameter can only have one lane going into it. If you want traffic to pass through it, you'll need to use single-directional lanes where only one of the connected lanes is going into the vertex.
2. The slotcar plugin doesn't have any special code path for docking. I'd be open to a PR for such a feature, but it's not something that we plan on having the core development team do since we're winding down development on this "old generation" Open-RMF while we ramp up development for the "next generation".

I recently published an interface proposal for the next generation of traffic management for Open-RMF. That proposal includes support for NURBS as trajectories anywhere along the robot's path. I think this will fit in much better with what you're looking for than the docking feature of the old generation. Feel free to leave feedback on that post if you think the proposal will be useful for your deployments, or if there's anything you'd like to see added or changed on it.