Harry van der Wolf <[email protected]> writes: > With the old route mechanism (A* and A* x2) I could tweak the routing > parameters by giving "massive priority" to toll roads that, despite > only travelling at set 90 kmh, it would still take the "route peage" > (with topspeed of 130 kmh). > Theoretically the shorter route over RN roads is faster. In practice, > it is much slower. I consider this a flaw in the routing mechanism and > priority setting. It is not simply "less kilometers per hour over a > shorter route = shorter travelling time" is better than "more > kilometers per hours over a longer, but much more optimized route". > Going with caravan or camper through small villages with traffic > lights and other hindrances takes in practice much more time.
There are two separate things going on and I think you may be blurring them. One is the search algorithm to choose among route candidates. There is a right answer and the question is the cpu cost of finding a route and then how how close that route's metric is to the best metric value. The second, that I have mentioned before, is that for a given proposed route (a list of how to drive from here to there), one must compute a metric. In theory, that metric (most used) is the time it will take. You are esentially saying that these time estimates are wrong. What's needed is to be able to output the route in great detail and then record what happens so that one can see how the reality lines up with the estimate. It could be that the time cost of stoplights is underestimated. It could be that the time cost of stoplights in your vehicle is higher than in a normal car. The relationship between maxspeed (capped at vehicle speed) and actually achieved speed may be off. All of this is tricky. I will repeat my usual plea: Do not use the word "penalty" to describe anything that is fundamentally about producing a correct travel time estimate. If a traffic_signal results in an average delay of 90s compared to driving straight through at speed, then adding in 90s is not a penalty. It's an estimate for the time to drive the route. If on the other hand you the human say that you do not like traffic lights, and you prefer route A that has 8 lights to route B that has 9 lights, as long as A takes not more than 2 minutes longer, then that is fair to call penalty, as you move from "time to travel" to "time to travel adjusted for a fake time cost for things I don't like". Personally, I often choose routes without scary left turns (we're drive on right), or other intersections that feel dangerous. I avoid roads that will have congested exit ramps (link roads) and especially backups before them. A crash is very costly in terms of time, both in the moment, repairs, etc. so I lean to "min probabilty of crash routing". -- You received this message because you are subscribed to the Google Groups "OsmAnd" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion visit https://groups.google.com/d/msgid/osmand/rmia4wdw52k.fsf%40s1.lexort.com.
