|
fishmech posted:Are you reporting this when you encounter it? I gave up on this about five years ago, after doing so extensively and not seeing the maps get fixed. E: also there are literally dozens of these intersections, and I'm not going through all of them on Google's behalf.
|
#
?
Jan 27, 2018 07:32
|
|
|
|
| # ? May 25, 2024 15:17 |
|
|
Lead out in cuffs posted:I gave up on this about five years ago, after doing so extensively and not seeing the maps get fixed. Yeah I fixed a local google maps problem like 5 times over a few years. Every time my change would eventually get approved by some higher ranked volunteer contributor, or I'd get grilled by the system about the change. Sometimes the update would go through but then months later I'd notice it was set back to being wrong again. I got big into doing google maps stuff but ended up being so frustrated by the system and my corrections all eventually falling back that I gave up.
|
|
#
?
Jan 27, 2018 08:35
|
|
|
Baronjutter posted:Yeah I fixed a local google maps problem like 5 times over a few years. Every time my change would eventually get approved by some higher ranked volunteer contributor, or I'd get grilled by the system about the change. Sometimes the update would go through but then months later I'd notice it was set back to being wrong again. I got big into doing google maps stuff but ended up being so frustrated by the system and my corrections all eventually falling back that I gave up. Fixing open street maps is also annoying as gently caress. If you want the community to fix things, make it easy.
|
|
#
?
Jan 27, 2018 09:03
|
|
|
nm posted:I literally tried to set a hospital as temporarily closed during the santa rosa wildfires with an article saying it was closed, and it kept getting reverted. That's like actually a problem is someone goes there with an emergency. Yeah I had a doctor's office in a huge strip mall thing with a weird parking lot divided into two chunks accessed via two different streets and if you went to the wrong one you'd have to drive around a big suburban superblock to get to the other. Google told me this doctor's office was in a little detached building on the far east of the complex with its own smaller parking lot, it's not, it's actually on the far west. I fixed it, it went through. Next time I needed to go I googled to make sure I remembered the directions and saw the location had changed back, I fixed it again and when I went I even asked if they had ever been at that other location, nope they've always been there and they get people coming in late all the time because google sent them to the wrong building via the wrong streets. I did this 2-3 more times, sometimes the change would stick around for a few months before resetting back. gently caress em.
|
|
#
?
Jan 27, 2018 09:15
|
|
|
Once GMaps registered a break in the main highway between Melbourne and Sydney, so if you had set a path between it'd route people onto a gravel road around Jugiong or Coolac or one of those villages about the midpoint of the route. Most normal people would likely not be too put out because they just cruise along the highway as normal, but for people who go by their nav rather than what's in front of them, that might be a problem. So obviously I reported that poo poo and it seemed to have stuck.
|
|
#
?
Jan 27, 2018 13:11
|
|
|
Speaking of maps, I found out that TomTom, big map maker / satnav device producer / and so on and so on, and in practice a rather important competitor to Google Maps, made a bunch of their APIs public a couple months ago. https://developer.tomtom.com/ Everything in the 'Maps' section to the top right is free to use up till 5 requests per second and 2500 requests per day. I don't know about the other sections, those have way more specific use-cases though. Now, the maps they send you through their API are simple vector/raster tiles. If you actually want to raw data such as road geometries that make up the maps you need to pay enterprise-amounts of money for their full map suite. But, nonetheless, their public APIs could be fun to play around in if you're a developer who wants to do something with maps and routing. By the way, I know quite a lot about how the software-level of modern maps, routing, navigation works. I know it falls kinda outside the scope of this thread but if anyone wants to know stuff about that feel free to ask.
|
|
#
?
Jan 27, 2018 15:04
|
|
|
nm posted:I literally tried to set a hospital as temporarily closed during the santa rosa wildfires with an article saying it was closed, and it kept getting reverted. That's like actually a problem is someone goes there with an emergency. Waze editing is way better. Takes a while to get into the meaningful stuff but the more senior editors are very accessible.
|
|
#
?
Jan 27, 2018 15:15
|
|
|
Carbon dioxide posted:Speaking of maps, I found out that TomTom, big map maker / satnav device producer / and so on and so on, and in practice a rather important competitor to Google Maps, made a bunch of their APIs public a couple months ago. Yes. How does it decide, why is it usually right, and am I being re-routed to distribute load across the system sometimes?
|
|
#
?
Jan 27, 2018 17:12
|
|
|
Carbon dioxide posted:Speaking of maps, I found out that TomTom, big map maker / satnav device producer / and so on and so on, and in practice a rather important competitor to Google Maps, made a bunch of their APIs public a couple months ago. Unless you live in Massachusetts and your states road inventory file is currently imploding (see my last post), at least in the US you can just download all that GIS data from your states clearinghouse, you don't need the tomtom API.
|
|
#
?
Jan 27, 2018 18:27
|
|
|
Roki B posted:Yes. Okay, so routing is actually one of the classic computer science problems. The first thing that is being done by any routing algorithm is that the program loads the map as a mathematical graph: a collection of nodes (addresses, points of interest) and edges connecting them (roads and road segments). Next, each edge is given a "length". This is not strictly the length of the road segment, but it tends to be a combination of things. It's often the average time it takes to get through that road segment at a certain time of day. Often, they give highways a lower "length" because it's usually more efficient to take them, instead of going through all kinds of narrow windy roads. That's especially important for routing algorithms that have a special mode for large trucks. There's all kinds of other penalties you can think of, but to apply them well, you need to know a lot of details about roads, not just where they are and how long they are, but data on legal restrictions on those roads, on speed profiles, all that kind of things. So step one is always having a Really Good Map to generate the graph from. Secondly... well, the way it's done nowadays is complex and I'll get to that. But routing once started with Dijkstra's algorithm (1959). It's not too difficult, but this video explains it better than I can: https://www.youtube.com/watch?v=GazC3A4OQTE I think the video says at some point that this algorithm always gets you the 'shortest' route (assuming the graph is correct). But there's one big problem. Even for tiny little graphs like in that example, it needs to try many dozens of alternative. Dijkstra had a good idea - but it doesn't scale at all. So, about a decade later (1968), some smart people came up with a improvement on Dijkstra's algorithm called A*. I'll let another nice Computerphile video do the explanation. https://www.youtube.com/watch?v=ySN5Wnu88nE Please watch both vids if you aren't familiar with the algorithms. The rest of this post assumes you know them A* is way better and way faster than Dijkstra's. It scales reasonably well. However, it doesn't scale forever. A small state or country already has many many millions of roads and addresses. A* just can't deal with that. Let alone routing across an entire continent. Okay, so there are some obvious small improvements we can do. - Simply start the algorithm twice, once from the beginning and once from the end. Make sure that the second one treats one-way streets in reverse and have them meet in the middle. - Give every road on the map an internal 'level', where interstate/intercountry highways are the highest level and local cul-de-sacs the lowest level. Now, make sure that the continental-wide highway level does not have any 'islands'. If you do find islands, internally upgrade some regional roads so that the map 'considers' them highways. Then set up each half of the algorithm such that when changing road types, it can never go to a lower road level. So, starting from a local road, once the routing algorithm finds a collector road that goes into the right directions, don't go back to a local road. Once you end up on a highway, don't drop back to a regional road. Do the same from the other end and have them meet in the middle. This way, you can make sure the route doesn't include some small nowhere mountain road in the middle of a international trip just because it thinks that's shorter than the highway a couple kilometers away. This only works if you first 'fix' your map so that the highway grid is complete and always connected, though. Both of these changes make the routing just a bit easier, but neither are enough to make long-distance routing work. Unless you're okay with waiting hours for your route to be ready. So, the main solution that happens these days is graph pre-processing. I am familiar with one way to do this, there are probably others but I can't tell you anything about them. The way I know is contraction hierarchies. That page gets rather technical so let me try to explain it in simple terms. Basically, you grab a map and have an automated script calculate millions upon millions of relatively short routes, only 2 or 3 graph edges long. Each of these calculations is simple, because it can just ignore any part of the map more than a couple kilometers away from your start and end point. Every time it calculates a route, it takes the edges and nodes on that route, and contracts them into one single edge between the start of the route and the end. In the data system, that edge contains some metadata linking it to its original internals, but these internals aren't used anymore for now. The new edge is called a 'shortcut' from A to B but in reality contains the entire route in there. This process is repeated a whole bunch of times, so that a bunch of shortcuts get contracted into a larger shortcut and so on and so on. Once this preprocessing step is done, you have a graph which is relatively simple and small. In the end, the algorithm usually decides where the remaining nodes and edges are, but as a simple example, you could end up with a graph that only has one node per major city and an edge linking each. You can 'zoom in', unwrap one of those city nodes and find out it contains a handful of nodes and edges, one node for the city center and then one for every surrounding smaller city around it, connected by edges. Zoom in further, you start getting nodes for individual neighbourhoods. Even further, individual streets. Now, when you want to calculate a route from address A to address B, what you do is start with a small part of the original, uncontracted, map around A. You start routing from there using A*. After a few steps into the route, you will hit a node that is part of the first-level contraction. At that point you zoom out a level. You start A* from there, using the simpler graph, until you hit a node that is part of a second-level contraction, you zoom out again, and so on. Do the same thing from B, meet in the middle, and you got a route you can calculate in seconds even with a small embedded processor in your satnav. Okay, so we got a good routing algorithm now that we can use in our satnav devices and on our phones. But is it perfect? The answer is no, not anymore. A* was 'perfect', given a perfect map it comes up with the most efficient route. But the way the shortcuts are generated in preprocessing means that by definition, the routing algorithm has less information available. So, at best, you end up with a very-good-but-not-quite-perfect guess. There's another quite big problem with this process: preprocessing is heavy. You need a big huge server with hundreds of gigabytes of RAM, quite some CPU, and even then an optimized preprocessing algorithm takes several hours at the least. So you simply can't redo it for every time someone wants a route. On top of that, once you end up with the preprocessed map, it can only do one type of routing, based on the profile you gave the preprocessor. In other words, you can't use the same preprocessed map to calculate the fastest route with truck restrictions and also the fastest route for a regular car. And you also can't inject live traffic jam information into a contracted map anymore, that just doesn't work because all kinds of different roads are merged into one edge. So, for every different type of calculation you want to offer to your users, you need to run that slow, big, preprocessing step again, store the new contracted map, and use that. And you need to repeat all that every single time you want to put a map update (including new roads, for instance) live. This is quite a problem and after looking at Google Maps with knowledge of these algorithms for a couple months, I think I can make a guess on how they solve it: - First of all, Google tends to show three alternatives. There's the main, fastest route, and two other options. I don't know if you ever noticed, but it seems that every single time, alternative 1 has the first half of the route be different, while alternative 2 has the second half be different. So my guess is, they do a very simple trick. They just tell their routing algorithm: okay now try it again, "but you're not allowed to use the same route for the first couple miles." And then they'll run it, saying "you're not allowed to use the same route for the last couple miles." That's all. I think, that at some point in the past they'd change up both ends at once, but that would more often than not get you a weird route that just wiggles back and forth along the main route, so they changed that. - Secondly, once it calculates the routes, it unwraps them, puts them on an actual complete map, and then applies live traffic info to them to calculate an ETA based on current traffic (or, if not available, average traffic at this time of week). It simply gets that data because the google maps app sends Google the speed every device on the road is moving at, based on its GPS. ETA calculation is in fact a completely separate step from the actual routing itself. - Thirdly, it comes up with alternatives while you're driving. I think that what it does is periodically call the routing function based on your current position as starting position, and see if any new alternatives come up. If so, it calculates the ETA of them, and if any of them turn out to be faster than the current route, it'll suggest that. It's not really all that smart, it just does some guesses and if one turns out to be useful, it informs you. I strongly suspect that Google is using machine learning to make their algorithms smarter. On top of that, every single day, many vehicles need to take roughly the same route. Every day, loads of cars drive from the suburbs of a city to its business district, etc. All those routes have an identical latter half. It would be a bit ridiculous to recalculate it every time, so they can just cache that on their servers and give the cached version to you when you ask for that route. They could even use their data on the actual most-used routes to give that extra priority on their next map preprocessing run. I couldn't say how much of this is still experimental and how much of this they are doing in practice, though. As I said, Google is rather secretive about it. Your last question was if you're being rerouted to distribute load across the system. I don't know, but honestly, I don't think they actively do that. It's just that if the routing algorithm already comes up with alternatives, and already suggests the fastest alternative based on the current traffic situation to you, vehicles will be distributed evenly without any additional work. See, you arrive at an intersection, and going east around the city is 5 min faster, so your satnav tells you and a couple dozen other cars to go east. But it's rush hour, and these cars add to the traffic jam, making the west road around the city 5 min faster. So the next couple dozen cars are automatically routed around the west because that's actually faster at that time. Now the jam on the west grows, so people are routed east again, and so on and so on. Elendil004 posted:Unless you live in Massachusetts and your states road inventory file is currently imploding (see my last post), at least in the US you can just download all that GIS data from your states clearinghouse, you don't need the tomtom API. Yeah, GIS data is raw map data containing road geometries and stuff. It's what you need if you want to build your own routing algorithm or show your own custom built map view. In many countries, a bunch of that data is available from the government, but in other places it's in hands of private parties. TomTom's APIs don't give you any actual GIS data, it's much simpler than that. Just map tiles and 'prerendered' routes. As I said, their GIS data is part of their enterprise products. The APIs are nice for people who don't want to dive into the GIS format right away. Carbon dioxide fucked around with this message at 19:25 on Jan 27, 2018 |
|
#
?
Jan 27, 2018 19:16
|
|
|
That's rad, thanks!
|
|
#
?
Jan 29, 2018 01:49
|
|
|
Who needs emergency stopping lanes anyway... New wire barrier installation somewhere on a B or C road in rural Victoria, Australia. A shitload of wire barrier is being rolled out in median strips of dual carriageway highways in rural areas to stop sleeping drivers from crossing into oncoming traffic, this is just terrible planning, couldn't have put it a meter to the left?
|
|
#
?
Feb 1, 2018 06:13
|
|
|
drunkill posted:Who needs emergency stopping lanes anyway... How well does the wire end up holding up in its proper usage? To the planned-for sleepy drivers I mean.
|
|
#
?
Feb 1, 2018 06:22
|
|
|
Good enough. A car or truck can hit it and take out a dozen 'posts' and it still stays up, easy enough to repair although it can take a few months if it is rural. Most of the time they are further away from the road about a meter off the shoulder so there are still emergency lanes. Although the myth about it decapitating or amputating motorbike riders still gives them the 'cheesegrater' nickname. Stops them coming off and hitting a tree head on. This looks to be the same stuff, can check the related videos as well: https://youtu.be/-zGJYVe-qms
|
|
#
?
Feb 1, 2018 14:44
|
|
|
drunkill posted:Who needs emergency stopping lanes anyway... I'm not familiar with this exact system, but typically guardrail (and even more so, guard-wire) systems have very strict requirements about the grading and slopes around them. The idea is that a car striking them needs to behave in predictable ways that have been tested. For example, if you put a curb in front of normal w-beam guard rail, it tends to launch people over the guardrail. Guard-wire systems tend to have a LOT of slack when a vehicle strikes them compared to w-beam, so if you were retrofitting a system that previously had w-beam (typically not allowing obstructions ~3' behind them) then the guard-wire system would need to be closer to the road to be effective. We would typically try to keep one of the shoulders usable, though. Giving them the benefit of the doubt, maybe they're planning to come back and restripe the road for a narrow inside and wider outside shoulder.
|
|
#
?
Feb 1, 2018 15:10
|
|
|
Looks like a great way to make sliced motorcyclist.
|
|
#
?
Feb 1, 2018 17:53
|
|
|
Jeoh posted:Looks like a great way to make sliced motorcyclist. Interstate 4, the deadliest road in America, has this stuff and people die every day because of it. Some of the deadliest sections are the ones that haven't been converted to a traditional steel guard rail yet.
|
|
#
?
Feb 1, 2018 19:55
|
|
|
Isn't there an argument to be made that since motorcyclists are relatively few, the wires do more good than bad?
|
|
#
?
Feb 1, 2018 20:58
|
|
|
Well I don't think motorcyclists tend to do well when smashed against any other barrier types. 
|
|
#
?
Feb 1, 2018 21:42
|
|
|
The cheesecutter thing is an urban legend. Statistically, cable barriers are just as safe for motorcyclists as guard rails, not as safe as concrete jersey blocks, but safer than no barrier at all. They're also half the cost of jersey blocks and are much easier to repair. Cable barriers are a great addition to highways and significantly improve road safety for all users. Guard rails are useful for cliffs and other danger spots where ensuring a vehicle is halted immediately is crucial. Concrete blocks are good for densely built high-speed roadways without medians, where the inherent risk of the road type justifies the higher cost of deployment.
|
|
#
?
Feb 1, 2018 22:33
|
|
|
Hippie Hedgehog posted:Isn't there an argument to be made that since motorcyclists are relatively few, the wires do more good than bad? The average human life is worth a couple million in court (or exactly 300k under Florida law). Only takes a few lawsuits to start putting a dent in your DOT budget. You'd be surprised what a little lawsuit risk will do to justify the higher costs of metal rails/barriers. Disability lawsuits aren't cheap, and it's far easier to prove liability for limb dismemberment via cheesecutter than blunt trauma against a K-rail. Varance fucked around with this message at 07:29 on Feb 6, 2018 |
|
#
?
Feb 6, 2018 07:26
|
|
|
Carbon dioxide posted:In other words, you can't use the same preprocessed map to calculate the fastest route with truck restrictions and also the fastest route for a regular car. And you also can't inject live traffic jam information into a contracted map anymore, that just doesn't work because all kinds of different roads are merged into one edge. So, for every different type of calculation you want to offer to your users, you need to run that slow, big, preprocessing step again, store the new contracted map, and use that. And you need to repeat all that every single time you want to put a map update (including new roads, for instance) live. I haven’t really noticed this and I don’t think it’s a consistent programmed behaviour. Sometimes all the routes converge at one end despite decent alternatives, and on the end where they are different, there’s an obvious best choice, a decent alternate, and an absurd third option that is different just for the sake of being different.
|
|
#
?
Feb 6, 2018 10:03
|
|
|
Platystemon posted:I haven’t really noticed this and I don’t think it’s a consistent programmed behaviour. I think it's based on behavior of users that they record from people running their apps. See - the example of an alternate route where you leave the arterial, drive around a large cul-de-sac development, then return to the arterial. Route adds 2 minutes to your trip, do you want to consider this alternative? No thanks. But if you're capturing users on a bus route (school or transit) then that loop makes sense.
|
|
#
?
Feb 6, 2018 15:14
|
|
|
Soo, the town attempted to repave my road last summer. They (barely) managed to get in the base coat(?) of asphalt before it got too cold and they couldn't finish. I say base coat because the road surface is about 2-3 inches below all the sewers. This led to their hilarious solution of putting a little ramp around all the sewers, and hoping that would last through the winter.. The road surface has turned to poo poo again, it's super bumpy and generally degrading again. Is whatever they put down as a base not intended to be driven on for long periods of time? Am I doomed to living through another summer of the road being entirely torn up and replaced, or will they just mill the top lightly?
|
|
#
?
Feb 7, 2018 01:23
|
|
|
The road is below the sewer lines?? I'm really not getting my head around what's going on with this road and ramps.
|
|
#
?
Feb 7, 2018 01:31
|
|
|
Baronjutter posted:The road is below the sewer lines?? I'm really not getting my head around what's going on with this road and ramps. I think he means the manholes are a couple inches proud of the roadbed at present
|
|
#
?
Feb 7, 2018 01:46
|
|
|
Anyone see Freeways, a little traffic management sim on Steam? edit: it's awful. Elendil004 fucked around with this message at 02:12 on Feb 7, 2018 |
|
#
?
Feb 7, 2018 02:07
|
|
|
Qwijib0 posted:I think he means the manholes are a couple inches proud of the roadbed at present Yea, sorry I meant manholes.
|
|
#
?
Feb 7, 2018 03:10
|
|
|
devicenull posted:Yea, sorry I meant manholes. Oh poo poo yeah I've seen that, where all the in-street elements sit a few inches proud because they haven't put the top coat or what ever on. They'll usually temporarily pile some asphalt around them, or just actually finish the road properly and not plan to do it when weather might stop it.
|
|
#
?
Feb 7, 2018 03:20
|
|
|
devicenull posted:Soo, the town attempted to repave my road last summer. They (barely) managed to get in the base coat(?) of asphalt before it got too cold and they couldn't finish. The latter. There's two general ways that you can work on an existing road that's in poor condition: 1. Milling/Grinding and Resurfacing, aka Shave and Pave. They grind off 2 inches +/- and then put 2 inches back. This can usually be done pretty quickly, and the surface they leave behind is usually only exposed for a few days or a week. 2. Full Depth Reconstruction is where they take out all the asphalt (6-8 inches or more, typically), and typically the aggregate below, and re-build it from the subgrade up. Maybe they also install other things like new curbs, new underdrain, storm drain, that sort of stuff. They can either do the whole road, or just portions. From your description, it sounds like they did full depth, except they stopped after putting the asphalt base down. Like you said, the asphalt base was probably put down while it was colder than it should have been, so that they could get the roadway reopened for winter. Ideally after they put it down hot, it would be compacted while the asphalt is still really really hot. They probably messed that up due to inability to control the temperatures. When you do asphalt work in winter you have to make really sure that you're controlling temperatures, and at a certain point the ground is just too cold to be able to construct it to spec. I would bitch and complain and point out to your local DOT or whoever did the work that the asphalt base was put down incorrectly/poorly, and it needs to be corrected before they put the surface course down. They'll probably shave-and-pave the base to correct it, then put a final surface course down. I would guess they already know they need to do this, and didn't worry too much about half-assing that base course since they're just going to come back and re-do it. But the squeaky wheel gets the grease, so it doesn't hurt to email.
|
|
#
?
Feb 7, 2018 03:49
|
|
|
This was originally announced for certain back in the summer, but the resigning and renaming is finally going into effect. Essentially the current layout of the road numbering looked like this: I-95 North would end at the US 1 interchange and then the road would continue south as I-295 South, which you'd take down to I-195 East to proceed to the NJ Turnpike interchange to continue on the other section of I-95 North. This is of course a remnant of how I-95 was meant to be built as a freeway from a bit west of the US 1 interchange up to the turnpike much farther north, but was canceled. This however left no possible all-controlled-access connection from the existing I-95 on the Pennsylvania side which crossed into NJ and I-95 coming across the George Washington Bridge and down the Turnpike. But finally, the long-delayed interchange between the PA Turnpike and I-95 is nearing completion, which means I-95 will absolutely be routed onto that and across the Turnpike Extension in NJ to go up that turnpike and neatly create a continuous I-95. But the question remained - what shall the new number be for the stretch of I-95 in PA and NJ that would no longer be I-95 after the reroute? Various options were considered, and for a while it was "certain" that the solution would be to route I-195 all the way to where I-95 would connect to the PA Turnpike, which would also mean essentially lopping off the top 7 miles of 295 and having a bit of an awkward interchange to continue on 195. This would after all work fairly well for the signage, because you could continue signing all of I-195 as east-west without too much hassle. But when push came to shove it was decided that the easiest way to handle this all would be to make all of the obsoleted I-95 section into I-295. Even though this will mean the NJ segment will now have I-295 North heading very clearly south for 3 and a half miles as is, it's not really that much more confusing than how 295 North became 95 South just by heading forward. And the PA segment at least will be signed East/West which is more or less true. Some proposals had been made to re-sign all of the existing and future I-295 length as "Inner" and "Outer" like is done with indicating things on proper beltways, but 295 won't really form a beltway so that was dropped. Here's the final plan.  And incidentally, here's the current route of I-295 from before the decision to re-sign and extend it: 
|
|
#
?
Feb 7, 2018 05:37
|
|
|
that sure is a thing: http://www.modot.org/kansascity/major_projects/Route_50_291_Interchange.htm https://www.youtube.com/watch?v=rW6eoyrRSfI
|
|
#
?
Feb 13, 2018 04:40
|
|
|
will_colorado posted:that sure is a thing: It's a diverging diamond with a roundabout for the northern intersection. It seems like you're creating a problem for yourself with the roundabout half, since the major traffic volumes have to cross over each other. Roundabouts work best when you don't have such unbalanced movements crossing each other up. Also, I hope people don't get confused driving on the left entering the roundabout.
|
|
#
?
Feb 13, 2018 15:13
|
|
|
Devor posted:Also, I hope people don't get confused driving on the left entering the roundabout. I think the geometry is such that it will present like a normal roundabout and nobody should get confused (other than your typical Midwesterner roundabout confusion!) 
|
|
#
?
Feb 13, 2018 17:09
|
|
|
nimper posted:I think the geometry is such that it will present like a normal roundabout and nobody should get confused (other than your typical Midwesterner roundabout confusion!) Someone traveling left on the mainline entering that roundabout will be confronted with people exiting the roundabout on the mainline, who in a 'normal' roundabout would be continuing on the circle. So your normal "I'm going to merge into the roundabout traffic" is actually a t-bone movement. It's not the worst it could be, but it's not conventional. Normal roundabouts almost eliminate t-bone movements where traffic is moving perpendicular - at worst you have a sort of oblique crash where someone in a left-hand approach lane, aiming to enter the inner radius, could hit someone traveling in the outer radius. This roundabout violates that expectation. Not sure how else you would accommodate those extra roads on the left side, though. Would definitely clutter up any configuration.
|
|
#
?
Feb 13, 2018 17:28
|
|
|
That thing is an eldritch horror.
|
|
#
?
Feb 13, 2018 19:12
|
|
|
|
|
#
?
Feb 14, 2018 03:21
|
|
|
nimper posted:I think the geometry is such that it will present like a normal roundabout and nobody should get confused (other than your typical Midwesterner roundabout confusion!) They should've done something like this instead. 
|
|
#
?
Feb 14, 2018 03:29
|
|
|
nimper posted:I think the geometry is such that it will present like a normal roundabout and nobody should get confused (other than your typical Midwesterner roundabout confusion!) That is super creative, I give them that. I got a headache when I thought about how you'd sign it.
|
|
#
?
Feb 14, 2018 16:52
|
|
|
|
| # ? May 25, 2024 15:17 |
|
|
no go on Quiznos posted:They should've done something like this instead. Look at this foolish person. Where are the reversing lanes, where are the roundabouts. In short, where is the heart. All I see are two standard signalized intersections! Putting all the interchange movements into one signalized intersection is going to make it operate like shiiiiiit peepeepants posted:That is super creative, I give them that. I got a headache when I thought about how you'd sign it. Just keeping putting up R15-8's until you're comfortable with the crash rate 
|
|
#
?
Feb 14, 2018 16:59
|
|