Hi! We see you’re using an ad-blocker. We’re fine with that and won’t stop you visiting the site.
But as we’re losing ad-revenue from this then why not make a donation towards website running costs?. Or you could disable your ad-blocker for this site. We think you’ll find our adverts are not overbearing!
Joined: Aug 25, 2004 Posts: 634 Location: Lincolnshire, England
Posted: Sat Jan 07, 2006 4:49 pm Post subject: COPILOT 6 ROUTING ANOMALIES
I thought this subject warranted a separate thread as route calculation has been a problem in past versions of Copilot.
What led me to post this today was a comparison of the route calculation speed that I was involved in over on the TomTom thread, and on another thread on this forum. It turns out that on a test route from City Centre, Spalding Lincolnshire to City Centre, Aberdeen, CP 6's Shortest route, avoiding tolls, makes a detour from shortest to go through the centre of Newcastle upon Tyne. This is a detour through a major city, lengthening the route by deviating from a direct route on major roads.
Detours such as these are not acceptable. Route calculation is fundamental to applications of Copilot's genre, so urgent attention really is needed to complete the job begun Copilot's routing, but plainly not completed. _________________ Dell Axims X50v & X30 (both WM 2003 2nd Edition). Copilot BTGPS3 and Fortuna Clip-on BT receivers. Jabra BT 250 audio headsets. Welltech 40032/AF32 BT handsfree audio. Copilot 6.0.4.110. TomTom Navigator 5.21. MS Autoroute/Pocket Streets 2005
Without dismissing your concerns PONDEROUS I don't like your chances of getting a quick resolution on this as it is a highly complex problem. There is probably no software in the world that will get every route right every time in a reasonable time. However, if the routing is substantially incorrect and consistently so then there is a case for a fix.
I analysed the specific route you mentioned, running on the desktop with the factory default speeds and neutral weighting with the following results:
Quickest route goes via the west coast; 493.8 miles; 9h 23m
Shortest route goes via the east coast; 428.6 miles; 9h 35m
Adding a waypoint on the A68 for the true shortest route gives 423.1 miles; 9h 38m
This represents just a 1.3% difference in distance. The reason for the discrepancy is undoubtedly due to the Copilot designers opting for speed of calculation over greater accuracy. But 1.3% could hardly be considered excessive.
Looking further though I tried to identify a worst case on the same route which turned out to be Low Conscliffe on the A1 to Otterburn on the A68. For this case the results were:
Calculated shortest route via A1: 66.1 miles; 84 min
"True" shortest route via A68: 58.8 miles; 82 min
This is a difference of 7.3 miles or 12%. The cost impact on the user is probably a quid or so.
So the question to consider is what level of inaccuracy can be considered unacceptable. Over 10% would probably be regarded as such so this could be a case for raising a bug report if this sort of result was found consistently on a large number of routes.
To illustrate the complexity of the software I did some more analysis choosing a route from Darlington (off the A1) to Rochester on the A68. I selected different destinations successively until I found two points close together: one of which routes via the A1 through Newcastle and the other that routes via the A68. The trip file can be dowloaded here. Load it onto desktop Copilot and see what happens when either the waypoint or destination are deleted and the route calculated. Also look how close the points are - about ten feet! The routing difference is about 4 miles or 6%. _________________ --Ken
HP iPAQ h2210 ROM 1.10.07 ENG
Windows CE 4.20
Sandisk Ultra II SD 512Mb
Haicom 303MMF and BT-401 slipper
Copilot Live 6.0.0.68
Joined: May 09, 2005 Posts: 283 Location: Reading, UK
Posted: Sun Jan 08, 2006 8:44 pm Post subject:
I adjusted my profile from the defaults to show a medium preference for motorways, and an increased average speed on motorways, and that tends to choose bypasses rather than town centres. At least with CP6, you can change that. _________________ Ian
iPaq 2210
Globalsat BT-338
Seidio G2500 Amplified Vent Mount
CoPilot 6, GPS Tuner 4.2, Navio 3.01
BMW 330ci Sport
Joined: Aug 25, 2004 Posts: 634 Location: Lincolnshire, England
Posted: Sun Jan 08, 2006 10:11 pm Post subject:
Ken S writes:
Quote:
I don't like your chances of getting a quick resolution on this as it is a highly complex problem.
Ken clearly speaks from a much greater technical knowledge than I do. Nevertheless it is still pretty clear to me that CP 6's routing problems, and the much greater ones seen in CP 5, arise because the routing logic works, and is adjusted, by reducing or increasing the use of the specified road types as well as taking into account the relative distances. I think that needs a re-think. It also makes the Advanced settings (road type preferences etc) somewhat hit-and-miss. For example, it won't really exclude motorways even if set to do so.
While accepting ianKB writes about adjusting the algorithm, I am not convinced that there is a set of universally usable settings, and adjusting the settings for each route is not in my experience viable.
It has certainly been possible to improve the routing far more than I ever thought possible. However, I still think that means need to be sought to eliminate these absurd deviations. Whether in the long term it will be possible to have both speed and accuracy of calculation, Ken does not come out and say for certain, but I accept his advice that it's not going to be ironed out overnight.
What will be needed in the short term (i.e. in the next patch), given these shortcomings are the Jump-to and other route and map viewing facilities that I mention on the thread entitled The Advantages of Copilot. With such facilties Copilot will stand head and shoulders above the competition DESPITE these routing irritations. However, no less is required. These facilities are not in my view optional extras or nice little touches, but missing essentials, especially while the routing needs some more work. _________________ Dell Axims X50v & X30 (both WM 2003 2nd Edition). Copilot BTGPS3 and Fortuna Clip-on BT receivers. Jabra BT 250 audio headsets. Welltech 40032/AF32 BT handsfree audio. Copilot 6.0.4.110. TomTom Navigator 5.21. MS Autoroute/Pocket Streets 2005
It has certainly been possible to improve the routing far more than I ever thought possible. However, I still think that means need to be sought to eliminate these absurd deviations. Whether in the long term it will be possible to have both speed and accuracy of calculation, Ken does not come out and say for certain, but I accept his advice that it's not going to be ironed out overnight.
The routing has improved considerably since CP5 as a test of the old culprits shows (Southampton-Birmingham, Southampton-Hastings, London-Dublin). Undoubtedly as computer price/performance continues to fall we will see further improvement. Still, no reason to just wait where there are genuine, systematic issues. However, we will need a large body of evidence to help the folks at ALK to track down the problem - or as I see it - to further tune the algorithm for UK conditions. What they need is plenty of information to identify trends but we need to be realistic in our expectations given that no software of this sort will give a perfect result every time.
Today I checked the shortest route from Whitchurch, Hampshire to Birmingham as a subset of the Southampton-Birmingham route which still looks a little suspect. As with my test above depending on ones start/end point the program may select different routes even if they choices are close together. Compared with a waypointed route determined by eyeballing the default route was 12 miles longer then the "ideal" 102 miles or 12%.
CP6 still shows a routing bias towards motorways as it does in the example above and as it did excessively in CP5. But is 12%/12 miles variation unacceptable? So far, in CP6, this is as bad as it gets. My question to the forum is what is an acceptable level of deviation in miles and/or percentage for Shortest routing. For my part I get uneasy when it exceeds either 10% or 5 miles. But then again I never use Shortest routing and if it were to take 5 minutes longer to calculate something better then I would probably live with it.
If we can agree on some threshold then as members find shortest routings that exceed these parameters we should collect together the saved trip files containing start, end and corrective waypoints and forward to ALK when there is a clear trend. That is probably the best way forward on this. _________________ --Ken
HP iPAQ h2210 ROM 1.10.07 ENG
Windows CE 4.20
Sandisk Ultra II SD 512Mb
Haicom 303MMF and BT-401 slipper
Copilot Live 6.0.0.68
I've been going back through some of the routes that used to cause problems for CP5. The Scraptoft to Rearsby route now seems ok as does the Waterford Tourist Office to Woodlands Hotel. One route that is still a BIG problem though is the Basingstoke to Coventry trip.
Quickest: 131.7 miles 2hrs 13min
Shortest: 124 miles 2hrs 41min
With waypoint on A34: 93 miles 2hr 14min
Thats a 33% (31 miles) deviation from shortest. Oddly, any other destination close to Coventry generates a more sensible route. Also, the shortest route runs along the A30 parallel to the M3 before joining the M25. _________________ --Ken
HP iPAQ h2210 ROM 1.10.07 ENG
Windows CE 4.20
Sandisk Ultra II SD 512Mb
Haicom 303MMF and BT-401 slipper
Copilot Live 6.0.0.68
I've now checked (I believe) all the problem routes reported from CP5 and nearly all of them now appear to work in CP6. The exception is those that involve the A34 between the M3 and M4 where discrepancies in the order of 30% are common. I have devised one that is a real mess between Durnley and Ardley. Interestingly the problem only occurs for northbound routes. The same route southbound is as one would expect. I am starting to suspect something to do with the map data for the northbound carriageway that is throwing the routing algorithm out.
Has anyone found a route (shortest) that is significantly longer than expected that does not involve the A34? _________________ --Ken
HP iPAQ h2210 ROM 1.10.07 ENG
Windows CE 4.20
Sandisk Ultra II SD 512Mb
Haicom 303MMF and BT-401 slipper
Copilot Live 6.0.0.68
Joined: Nov 04, 2005 Posts: 18 Location: Crewe, UK
Posted: Mon Jan 16, 2006 9:56 am Post subject:
Has anyone checked any routing in France?
Leaving aside the common problem in CP5 & CP6 of routing through traffic-calmed town centres rather than using the bypass, there now seems to be a new problem with the A11 between Le Mans and Angers: It seems to be impossible to leave the motorway except at Junction 14 at Angers. For example, if you try and force to leave at Junction 10 with a waypoint, it routes you down the motorway to Junction 14 then back up the Route Departmentale parallel to the motorway: about a 50-mile detour . The slip roads are shown on the map, but don't seem to be accessible from the A11.
While I can't comment on France in particular (I only have the UK version), but I have had similar problems when 'forcing' a route using waypoints in the UK.
CoPilot can be quite (very!) particular when it comes to placing waypoints. I have had significant problems with intersections such as motorway junctions. When I have wanted to use a specific junction, I have generally found that placing the waypoint at a particular junction, at best hit-and-miss, and at worst a complete waste of time. This has also included enlarging the map and placing the waypoint at various positions on slip roads within the junction itself.
I have given up trying to place waypoints on junctions, what I do now, is to place the waypoint a little way up the road I want to join, making sure that I am clear of any junction. It is also essential to ensure that where the road is held as a dual carriageway, that the 'correct' carriageway is selected, otherwise you will be taken past your waypoint and back again.
In my experience with placing waypoints, the one thing which seems to have caused me the most problems, is where the waypoint is placed in the central reservation - this should be avoided.
I am of course assuming that you have taken the side of the road you drive on changes between the UK and France into account.
Joined: Nov 24, 2003 Posts: 1441 Location: Swansea
Posted: Mon Jan 16, 2006 12:09 pm Post subject:
redhorse wrote:
Has anyone checked any routing in France?
Leaving aside the common problem in CP5 & CP6 of routing through traffic-calmed town centres rather than using the bypass, there now seems to be a new problem with the A11 between Le Mans and Angers: It seems to be impossible to leave the motorway except at Junction 14 at Angers. For example, if you try and force to leave at Junction 10 with a waypoint, it routes you down the motorway to Junction 14 then back up the Route Departmentale parallel to the motorway: about a 50-mile detour . The slip roads are shown on the map, but don't seem to be accessible from the A11.
nick
I started a thread a couple of months ago about French routing. In my case, I found that it was impossible to leave the A26 at Junction 11 when travelling south. I was careful to zoom right in so that I could see both carriageways, and to pick the correct one for driving in France! I got exactly the same problem planning a journey on the desktop - it would route perfectly as far as J11, then if I started a new route just after leaving the A26, on the D1, it would route perfectly well from then on.
I tried quite a few other routes and didn't have the same problem - it's disappointing to hear that something similar happens on the A11
Robin _________________ TomTom One v3 Europe, Navcore 7.903, Western Europe 835.2420
TomTom Go 720 Europe, Navcore 8.351, Western Europe 855.2884
Satmap Active 10
Plus a lot of other PDA GPS kit, seldom used
I think you are right re A34. CPL4 had no problems and routed fine Northbound on the A34. CPL5 wouldn't voluntarily and neither will CPL6 and seems even more bizarre.
My hunch has always been it's not the A34 between the M3 and the M4 that's the issue but the A34 between the M4 and Oxford.
Certainly with CPL5 I would be routed up to the M4 on the A34 but then nothing short of a Via would convince CPL5 to take me North at the A34/M4 interchange.
CPL6 is worse for me as it now wants me to go M3/M25/M40 if I'm heading to Warwick/Coventry etc.
Joined: Aug 25, 2004 Posts: 634 Location: Lincolnshire, England
Posted: Mon Jan 16, 2006 2:49 pm Post subject:
KenS writes
Quote:
... we will need a large body of evidence to help the folks at ALK to track down the problem - or as I see it - to further tune the algorithm for UK conditions.
That's true, and it is clear that ALK have listened on the question of routing. They do need to listen, or at least act upon, some other related points, though. While, for example, Neil01's present analysis is new to the forum, the fact that there are problems with waypoints is something that has long been common knowledge.
It as also been common knowledge that CP 6 lacks comprehensive, or even basic, route review facilities. These are essential to tide us over until there is a Copilot version that consistently produces fair approximations of the type of route desired. It is simply no good having the means to insert waypoints, or to tweak the routing, if comparing the results takes a prohibitively long time. I know this is repetition, but it is worth saying again, I believe. _________________ Dell Axims X50v & X30 (both WM 2003 2nd Edition). Copilot BTGPS3 and Fortuna Clip-on BT receivers. Jabra BT 250 audio headsets. Welltech 40032/AF32 BT handsfree audio. Copilot 6.0.4.110. TomTom Navigator 5.21. MS Autoroute/Pocket Streets 2005
Joined: Nov 04, 2005 Posts: 18 Location: Crewe, UK
Posted: Mon Jan 16, 2006 3:04 pm Post subject:
My waypoint for the 50-mile detour was positioned at a roundabout just off the A11 J10. I also tried putting a waypoint on the exit slip itself (allowing for driving on the right!), which can be clearly seen at full zoom. CP6 said it was impossible to make a route to it. Didn't make any difference whether I used my Mio168 or my PC.
France has always been a problem, as CP5 and CP6 seem to regard all non-motorways as 30 mph (or less) roads, whether dual carriageway or not. Ouistreham port to the Caen Perpherique it routes you down the small road to the East of the Orne canal, rather than down the fast dual carriageway on the West bank.
My last trip to France taught me more about placing waypoints in CP5 than I wanted to know.
...I also tried putting a waypoint on the exit slip itself (allowing for driving on the right!), which can be clearly seen at full zoom...
I have also found through experience not to place waypoints on sliproads to/from roundabouts or on the roundabout itself. I would try placing the waypoint a little way down the road you wish to take.
I now avoid placing waypoints on or very near to a 'slip road' of any description. I always try to keep to a plain stretch of road. Since following this simple 'rule' I rarely have any problems, and so far, they have been as a result of map errors.
Posted: Today Post subject: Pocket GPS Advertising
We see you’re using an ad-blocker. We’re fine with that and won’t stop you visiting the site.
Have you considered making a donation towards website running costs?. Or you could disable your ad-blocker for this site. We think you’ll find our adverts are not overbearing!
All times are GMT + 1 Hour Goto page 1, 2, 3, 4Next
Page 1 of 4
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
Or you could disable your ad-blocker for this site. We think you’ll find our adverts are not overbearing!
Hi! We see you’re using an ad-blocker. We’re fine with that and won’t stop you visiting the site.
But as we’re losing ad-revenue from this then why not make a donation towards website running costs?. Or you could disable your ad-blocker for this site. We think you’ll find our adverts are not overbearing!
Forum FAQ
Search
Usergroups
Profile
Log in for private messages
Log in
