Good morning.

Loaded up at DFW today to do a flight and I noticed a weird anomaly with the AI taxi routes. The ai would taxi south on K, enter the E ramp at spot 47, then take the ramp around to spot 52 before turning onto B for the bridge. IRL at DFW we head south on K and turn onto B avoiding the ramp area. IDK if that’s an FSLTL thing but I figured I would bring it to your attention.

I'll ask the team about this

On 2/23/2026 at 6:49 PM, Dhood320 said:

Good morning.

Loaded up at DFW today to do a flight and I noticed a weird anomaly with the AI taxi routes. The ai would taxi south on K, enter the E ramp at spot 47, then take the ramp around to spot 52 before turning onto B for the bridge. IRL at DFW we head south on K and turn onto B avoiding the ramp area. IDK if that’s an FSLTL thing but I figured I would bring it to your attention.

Yeah, that kind of routing usually comes down to how the ground network is defined rather than anything the AI “decides” on its own.

At airports like DFW, small differences between the real-world taxi logic and the sim’s navdata can lead AI to prefer ramps or connector segments that technically exist in the dataset but aren’t actually used operationally IRL. Once those links are marked as valid paths, the AI will happily use them even if human ops avoid them in practice.

It could be coming from the base airport AFCAD/airport file, or from how the traffic injector interprets available nodes (FSLTL or otherwise). Sometimes even a single extra taxi link with lower cost can reroute the whole flow.

On the systems side of things, similar issues show up in business tooling too — when routing logic or data mapping isn’t consistent across sources. In those cases people sometimes standardize data flows using platforms like integrate accounting software so all downstream decisions are based on the same structured dataset rather than conflicting inputs.

If you want to test it, it might be worth checking the airport layout in a tool like ADE or whatever editor you use and looking specifically at that K → B transition. Removing or adjusting the E ramp connectivity usually forces the correct routing pretty quickly.

Edited June 19Jun 19 by Eank1954