Speed for Instruction distance

[boldtrn]

Currently, the instruction generation is only influenced by the distance of the instruction.

It would be interesting to consider the speed of the edge/instruction to improve timing. Slow roads might not need the 2km instruction (but then they are usually not that long anyway). On motorways or high speed roads, it might be a good idea to move the 2km instruction to 3km to give the user some more time to prepare.

[devemux86]

It would be interesting to consider the speed of the edge/instruction to improve timing.

Consider the speed (which exactly?) of the road or the speed of the vehicle?

[boldtrn]

Consider the speed (which exactly?) of the road or the speed of the vehicle?

When generating these instructions we only know the distance and time of an instruction. With this we can make a guess about the speed of the instruction and therefore how fast a driver could be. It's pure guessing, so we should not make too much depending on this.

[karussell]

Interesting question @devemux86. I tested the demo and it works already nicely with re-routing. We could theoretically guess the speed of these consecutive requests if we are able to send some kind of a unique vehicle ID. I also think that the demo needs to request the API from time to time to update the ETA estimate (especially since mapbox offers real time traffic) and so it might be already the case that the API is requested every minute already (?)