This morning I drove 60 miles. I had a choice of two routes: one is slightly longer but has more reliable traffic flow, while the other is shorter but with highly variable traffic conditions. Knowing of this, I opted for the former, despite Google Maps claiming the latter would have been one minute faster. My assumption was that Google was basing its routing on current traffic conditions rather than statistical expectations of traffic conditions at the time I would have arrived in the high traffic zone.
Am I wrong about that? Is Google Map’s routing software smart enough to predict what traffic conditions are likely to be when I arrive at point X (based on its historical observations of traffic-versus-time at point X), and base its routing recommendations on that? Or does it just recommend a route based on what current traffic conditions are (before you depart) at all points along the suggested routes?
Thanks for the link. From that page I see now that it’s possible to examine typical traffic anywhere for an arbitrary time of day. This means Google is keeping track of typical traffic levels versus time; it’s not explicitly clear that it’s using that traffic database (rather than current traffic levels) to choose routes, but I guess it seems likely.
Google sends traffic alerts to my Android phone about traffic on my route to and from work all the time. I look at my phone when I leave work and with one swipe see my estimated time to home. Same thing in the morning, the estimated time to work is just there. Sometimes it’s suspiciously long and sure enough, it will be a heavy traffic day.
I tend to use Google maps to navigate to and from work every day. It usually tells me what traffic levels are like (heavy, normal, light), which implies to me that they’re comparing current conditions with their database of traffic patterns.
I also suspect that there’s probably some sort of time/distance window such that if you’re inside it, it’ll use the current conditions for routing, and outside of it, it’ll route you based on the historical patterns.
So if you’re driving from say… Dallas to Houston, it doesn’t make a lot of sense to try and prognosticate traffic patterns 5 hours in the future based on current conditions, but the historical database is probably extremely useful for that task.
Similarly, if you’re going 10 miles, like I do to and from work, current conditions are probably much more immediately pertinent to that trip than historical conditions.
Their traffic routing is still not perfect; they always tell me to take a particular road and turn onto a freeway access road, without realizing that it’s ALWAYS backed up past the road, and it’s always a slow motion beating to turn right and fight across 3 lanes of traffic to turn left in 250 yards. It’s actually faster to drive up a block and turn right upstream of the traffic and get in the right lane initially, but Google Maps (or Waze) can’t model that well.
Also, there are some times that a bit of local knowledge trumps the computer; typically it’ll have me enter the freeway andthen drive me through the middle of a traffic jam, as that’s the calculated fastest route. But that’s not always so; getting off the freeway and taking the access road all the way down is frequently faster because again, they don’t calculate the time it takes to change lanes across several lanes of congested traffic to stay on the freeway.
I often drive 45 minutes to work downtown in the early morning. Normally I’m ahead of most of the traffic but things are thickening without much slowing (yet) by the time / place I arrive.
But if I leave later than usual I’ll get into the exponential delay curve where I’m hitting each point in the growing clog later and later so it’s cloggeder and cloggeder. IOW, leave 15 minutes late and arrive 35 minutes late. We all know the experience.
Plain old Google maps on my Android seems to compute travel time based only on current conditions throughout the route. Not the conditions that will be expected at each point when I’m expected to get there.
Some other map apps might be smarter.
I live in the middle of nowhere, where it usually doesn’t matter, but on a recent training trip to Denver I was grateful that my Google Maps navigated me around high-traffic areas, even changing my route on the third day based on the current traffic conditions.
I drive 20 miles thru rush hour traffic just north of Detroit everyday and Google Maps is a gift from god. Very accurate in predicting time to arrival and even let’s you know if conditions change while you are driving and gives alternate routes. I generally have about four options to get across town to work and back, but sometimes it will give me a fifth option that has me cutting thru a neighborhood around a closed bridge. I even used it on a trip from Detroit to South Padre Island Texas and distance doesn’t seem to be a factor, it still calculates based upon construction traffic and accidents. I’ve stopped second guessing the app a long time ago and just go with the flow.
Google Maps (version 9.27.2) on my Android phone (Samsung Galaxy S7) calculates the trip time based on current traffic conditions and updates it in near-real-time while I’m driving.
Sometimes, while I’m following her directions, The Lady will suddenly alert me that there’s a major slow-down (or that there’s been an accident) on highway X ahead and that I could avoid it by using this new route instead (tap here to change). It’s saved me several minutes, many times.
Yes. This is, in fact, pretty standard in the navigation industry right now. The algorithmic specifics are, as you might imagine, trade secrets. But in broad strokes, every company that offers real-time traffic monitoring also has databases that help it predict what traffic will be doing later in the day, or next Thursday (provided the Earth doesn’t get bulldozed to put in a new hyperspace bypass). As real-time traffic data has a pretty short shelf life, a routing algorithm will start using those databases to its advantage pretty quickly. Regression to the mean and all that.
You can test this, by the way. Ask your favorite routing tool for a cross-country route that goes through a major city around rush hour, but that rush hour is a few hours in advance of your current time. Chances are you’ll see some congestion on your route through that city, some congestion that doesn’t resemble the current map of traffic congestion.
On the other hand, I once asked Google for bicycle directions to a particular point downtown that I go to often. I knew the last leg of the trip (from crossing the bridge to the end, at the least), but I wasn’t sure where the best places were to change roads before that, and I wanted to know the total traffic time.
I was briefly surprised that Google gave me exactly the same last leg of the trip that I would have taken myself, even though the best route I’ve found has a couple of unexpected turns. But then I realized that the population of people regularly traveling by bicycle to that point from the west side with an Android phone is fairly small… small enough that I myself make up a significant fraction of it. In other words, Google was giving me back the route that it had gotten from me in the first place.
Google usually tells me there’s a traffic jam on my route 15-20 minutes after I’ve encountered it. Though their real-time navigation app occasionally makes route changes during the trip to avoid accidents and standstill traffic, I’ve noticed.
I just don’t like the computer voice interrupting my music to tell me to turn when I already know how to get onto the highway from my house! I just need the exit for my destination, please. So I don’t use the navigation app that much.
A few years back, I was driving from here (near Washington DC) to Portland, Maine. Google really, really wanted me to take the route through New Jersey, NYC and Connectictut, claiming it would save me 2 hours versus going through PA and various routes that mostly avoided NYC.
So finally I listened. And got stuck in Connecticut - which talking to other people, this was expected and normal traffic for Friday afternoons. Ugh.