NB: you can change the endpoints graphically. Though it's gotten caught up on an unnecessary roundabout, as if it were a way-point (even retracing the route to include it...).
EDIT: Disregard. Found the elevation data:
I personally think it's fine.
The price (at least for me) was not a _frequently asked_ question, but rather the _very first_ question I thought of asking.
Love the elevation charts, really well done.
One route indicated the lane cove tunnel in Sydney, a place where I can't run or cycle, I could get squished.
Would be interesting to use long/lat co-ordinates, could be really great for bushwalkers and their specific location.
Combine this with something like zombie-run, have the zombies attack at the bottom of a hill/vice versa, you could have a hit with the hard core running community.
Would be nice to see the alternative routes, not just the flattest, but maybe challenging ones, best rated.
But what you've done in a few weeks is still awesome. Well done.
Also, any idea why this doesn't work for Berlin, Germany?
You could at least show the ZERO_RESULTS status somehow.
...it does route planning, but the speeds aren't just based on distance travelled but also the elevation. So, it's not so much that it avoids the hills, but it'll only send you over them if it's not faster to avoid them.
Cyclestreets has long talked about going open source - I don't know where they are with that - but there's some details of their routing algorithm here:
That page doesn't mention the elevation part, but they use Naismith's formula, and they give examples here:
Another example is going from "nanaimo and 12th, vancouver, BC" to "nanaimo and 15th, vancouver, BC": There's a bridge in the middle of that which crosses a 30' deep chasm, but you're showing it as being "very difficult" because you're calculating for someone cycling down the side of the chasm -- which would indeed be nearly impossible!
Seems like it would be a little bit tricky to try to systematically identify steep ravines like this in Google's elevation data, but it's certainly an interesting problem.
I see a routing problem. I had assumed this tool would give longer routes to avoid steep inclines, but that's not always the case. Take the suggested case: Nob Hill to Russian Hill in San Francisco. Drag marker A to the corner of Vallejo and Powell (to the east of that big hill.) Now drag B to Union and Polk (to the west, and one block north of Green.) The resulting route includes some "very difficult" parts along Broadway when it could suggest one block over to Pacific and eliminate those black dots.
This way, you could make tradeoffs between distance and effort.
I'd happily travel an extra flat mile to avoid a .25 mile hill with a 9% grade.
Note that while many truckers travel recurring routes,
some pick up ad-hoc LTL (less-than-truckload) shipments
and must replan routing with nearly each shipment. For
fun, demo using:
Tucson, AZ -> Tucumcari, NM
Tehachapi, CA -> Tonopah, NV
What's the basic idea behind the optimization algorithm?
It's a navigation system specially designed for bikes, it takes into account how flat is the rout among many another things like bicycle lanes, speed and safety, it's really quite cool.
Disclaimer:I work for them!
Having played around, the route given seems to vary slightly as I move an endpoint slightly at either end and then massively recalculate - perhaps it caches certain segments and favours those?
Just curious really, it's a cool idea!
PS My wife is in Hackbright as well, it is an exciting time in the program (for other HNers, it's the last week of the current cohort). I heard about FlattestRoute a few weeks back, it looks really great!
For example, Bolinas, CA to Vacaville, CA has steeper segments than the Bolinas, CA to Dixon, CA even though they take the exact same route out of Marin and Sonoma.
The extreme example, Bolinas, CA to Boston, MA never gets above a four percent slope.
There's a pre-existing tool I like to use, which is not nearly as nice looking but does allow you to specify your maximum grade: http://amarpai.com/bikemap/