Tuesday, July 12, 2016

3200 Miles with CoPilot RV: It's the Best of Bad Choices for RV Navigation

We just returned from a 3200 mile, 12-day RV vacation starting in and returning to Iowa and hitting the states of Illinois, Missouri, Oklahoma, Texas, New Mexico and Kansas. Here's our route:


Six years ago this week, we were on a nearly 4100 mile trip to the east and southeast US. On that trip, in 2010, I road tested ALK's CoPilot app for the first time. A thorough look at my experience with that app became one of my first substantial blog posts. As my now 6 year old blog post outlined, there was good, bad and ugly in my experience with CoPilot. For me, the bad and the ugly outweighed the good and I stopped using the app for many years.

Completely dissatisfied with the current choices available for RV navigation, I immediately spent $50 to purchase CoPilot RV the first day it was released this spring. I was willing to give the app another chance 6 years after my last CoPilot experience. On this trip, for all 3200 miles, I ran CoPilot RV on a Samsung Tab S2 8.0 tablet as well as Google Maps on a Samsung Galaxy Note 5 smartphone.  CoPilot was rock solid from a performance and stability perspective. In 3200 miles (48 hours of driving), the app never locked up, froze or crashed once.

Disclosure and About Me: I'm a bit of technology evangelist, both personally and professionally. I have a strong interest in navigation but have not worked professionally in the field. I am a volunteer beta tester for 3 navigation companies and am also a volunteer Google MapMaker Regional Lead. Over the last decade I have owned and used the following solutions: Garmin SP III, Garmin SP 2710, Garmin SP 2730, Garmin i5, Garmin c330, Garmin 760, Garmin 885T, Dash Navigation, TeleNav for Sprint, TomTom 740 Live, CoPilot Live v8, Navigon for Android, Navigon for iPad, Google Navigation, Waze, Garmin Nuvi 3490LMT, Garmin 3590LMT, TomTom for Android, Rand McNally 7725,  Rand McNally 7735, Garmin viago, Rand McNally RVND, CoPilot RV for Android.

CoPilot navigation apps have a long history across many different mobile platforms. ALK should be commended for believing in the smartphone navigation space and sticking with it through many years. In general, CoPilot apps are extremely feature-rich, accomplishing nearly every feature you'd find on a traditional dedicated navigation unit. The abundant features make CoPilot RV an attractive alternative for RV navigation. Notable features include:
  • Ability to pre-plan, save and recall routes. Multiple routes can be saved.
  • Routes can contain many stops and stops can be manually managed or automatically optimized.
  • User-installed points-of-interest (POI) databases are supported so content from sites like POI Factory can be installed.
  • User-entered RV dimensions and weight are used in route calculations.
Summary
I'll cut to the chase first and then provide more detail for those interested. The title of this blog post is "It's the best of bad choices for RV navigation." That summarizes my feelings about CoPilot RV. CoPilot has drastically improved in the key feature of route/ETA calculation since my experience in 2010. However, like was the case in 2010, CoPilot RV's map quality remains a significant downfall. Unlike after my 2010 trip with CoPilot, I will continue to use CoPilot RV but will be watching and hoping for other solutions due to the map quality issue. A RV navigation app carries an additional burden of having RV-specific attributes such as weight, height, size and propane restrictions. I'm unclear how ALK suggests customers should have confidence in the accuracy of these very important RV-specific attributes given the persistence map quality issues over the years.

The Details
This blog will not be an exhaustive look at CoPilot's features. My 2010 blog post provides this and actually, the look and feel of the app has remained similar to the 2010 app. Normally that would be a red flag but, despite CoPilot not abiding by any Google material design guidelines in any way, I find it visually pleasing and relatively easy to use. CoPilot does register as an intent handler within Android so clicking addresses within Contacts, Calendar, etc., can be handled by CoPilot automatically.

CoPilot's route/ETA calculation has been fixed from my 2010 experience and is now very good. However, users MUST buy the optional $9.95 annual traffic package in order to experience the improvement.  On this trip we traveled 6 - 11 hours per day on our drive days. I found the ETA that CoPilot calculated to be realistic for the total trip time (i.e., inclusive of gas stops). Comparatively, Google Maps' ETA calculation seemed to be more accurate for actual drive time, but would then add time for each gas stop.

My recommendation is simple, do not use CoPilot RV without buying the optional traffic subscription. I'll explain.

Without the optional traffic package, CoPilot has poor "knowledge" of road speeds. CoPilot appears to classify roads in groups of similar road types and then assigns a generic speed to each road type. The user, in a setting that has not changed since 2010, can input custom road speeds for each of the generic road categories, and also select the user's preference for driving on each road type. As you can read in my 2010 blog post, the result, even when using default settings, is disastrous route calculations and ETA times.

Here is an example of a route/ETA calculation in the current CoPilot RV app using the default app settings and not using the optional traffic package. From personal experience, a time of 8:43 is absurd for this route, as CoPilot itself will confirm once you actually activate the traffic package.

CoPilot Support will say "well, just change the default speed to something more realistic." Nope. The problem with that is CoPilot road categories don't distinguish between rural and urban roads when setting custom user speeds. So, if the user sets a divided highway custom speed to 50 mph for example, CoPilot applies that speed to a divided highway whether it runs through rural farm land or is an urban divided highway in the middle of a dense urban area.  If the user corrects the ETA in this rural route example by raising the custom speed, it may bring down the 8:43 ETA to something more realistic but the result will also drastically underestimate the ETA in an urban environment where the custom speed cannot be maintained on urban streets badged as a highway. Luckily, CoPilot has "fixed" this problem if you purchase the optional traffic package.


Here is the exact same route with the optional traffic package activated. An ETA of 6:42 is a very realistic calculation for this route. This optional feature saves CoPilot RV from being removed from my Android tablet, never to be used again, like occurred in 2010.


So what is going on here? The optional CoPilot traffic package actually includes two features. The first feature is branded ActiveTraffic, which is real-time traffic content. The second feature is ActiveRoutes, which is essentially historical traffic content. ALK describes it as follows: "ActiveRoutes calculates your route using actual road speeds at the specific time of day and day of week you're traveling."

This means that the generic road category and road category speeds are not used when ActiveRoutes is active. Instead, the route and route ETA is calculated using a massive (3rd party) database of knowledge about actual road speeds and how those speeds vary by time of day. This is how other modern smartphone navigation apps like Google Maps calculate routes and the ETA. This feature is vital to making CoPilot RV usable.

2D View
I believe that most people prefer to use navigation apps in 3D mode (the default view in Google Maps). However, I really enjoy CoPilot RV's 2D view using the "zoom to next turn" feature. Theoretically, this feature allows for a nice overview of long stretches of your trip while zooming into the details as you approach a turn. While the next turn remains far away, the 2D view is north up but switches to "direction up" as you get within a dozen or so miles of a turn. I find that the 2D view provides a great overview of your overall route and gives you an at-a-glance sense of your trip's progress.  Here is an example of the 2D view with a next turn far in the distance.


"Splits Off" Annoyance
I used the word "theoretically" in the paragraph above because in practice, this 2D view is interrupted by a flaw in CoPilot RV. I refer to the problem as the "splits off fixation." CoPilot constantly provides a meaningless instruction to "Stay on Road1 while Road2 splits off." And every time it provides this meaningless instruction, it defines a point on the map that CoPilot considers a "next turn." As a result, CoPilot RV's 2D display is always zooming closer and closer to these meaningless "splits off" points. I'd estimate that over the course of this 3200 mile trip that CoPilot provided a "splits off" instruction at least 100 times that I'd categorize as completely unnecessary.

Even worse, I have CoPilot set to provide three audible upcoming turn prompts as the next turn is approached. CoPilot uses these three prompt distances to announce these "splits off" warnings too. So 100 unnecessary uses of the "splits off" alert turns into 300 audible prompts.

The "splits off" alert has good intentions. It is a nice feature in the specific circumstance where a road legitimately "forks" and the driver could go multiple ways. However, CoPilot RV provides this "splits off" instruction on things like standard freeway exit ramps even when the ramp isn't one of the freeway drive lanes. It also repeatedly does it when two highways intersect, even when that intersection would require a right or left turn using a dedicated lane. Here is an example. Notice the intersecting Highway 1695.  It has absolutely nothing to do with my route. At Highway 1695, I-70 is elevated and Highway 1695 passes below. Highway 1695 is accessible via standard freeway entrance and exit ramps.  Despite this, CoPilot RV insists on warning me "Stay on Interstate 70 as Highway 1695 splits off." It's a ridiculous instruction and it occurs over and over again in the app.


Here's another example. It's the same scenario, this time with Highway 40.  I could have screen captured dozens and dozens of these situations.


So, we have this annoying "splits off" situation but yet there are situations where CoPilot is silent when there should be instructions provided. As I approached this small Kansas town on Highway 56, notice where the yellow arrow is. This arrow represents the "next turn" as far as CoPilot is concerned. It is 1.6 miles away from my current location.  The problem here is obvious. There should be a left turn instruction where Highway 56 meets "Old 56" on the map. This is a stop sign, not a curve in the street. CoPilot is silent in this important scenario yet chirps unnecessary "splits off" warnings where it isn't needed.


Here is another example of CoPilot not providing instructions when instructions are warrantied. This is another example where the trip required a stop and a right turn in order to stay on the highway but CoPilot provided no instructions at all.


Search Along Route is Broken
A key feature in any RV navigation app is a "search along route function" for finding the next truck stop, dump station or campground. CoPilot RV has a "search along route" feature but, at least in my usage, it's completely broken. It simple does not return any content, despite there unquestionably being content that should be returned. Here is the result of "search along route" for gas stations. The search was started on I-70 east of Topeka headed towards Kansas City (not yet to Lawrence). Gas stations are plentiful along this route. The round icon to the left of the "Map" button spins continuously. This screenshot was taken after letting the screen sit for at least a mile.


POI Inadequacy
CoPilot RV allows you to input your own custom points-of-interest databases from sites like POI Factory. It's a good thing this feature exists because I found the built in POI database to be outdated and paltry. We overnighted in the small town of Dodge City, KS. It also served as a RV gas stop. CoPilot RV says there are three gas stations in Dodge City. Nope.

 

In reality, there are over three times that many including a Love's and a Flying J truck stop.  Google Maps (correctly in my opinion) includes truck stops when searching for gas stations (but not vice versa obviously). CoPilot however, does not. The Love's and Flying J are not found in the gas station POI search in CoPilot but are found when searching the truck stop category. In my opinion, the truck stop category should be a subset of the gas station category. They should not be mutually exclusive.

Signpost Imagery - CoPilot Creative Marketing
ALK advertises signpost imagery as a feature of CoPilot RV. Signpost imagery is a great feature because it helps drivers associate the maneuvers shown on the screen with what they are seeing in real life. Combined with lane guidance, these are meaningful features for RV drivers since quick, last minute lane changes are hard to make in a RV. These features make big city driving easier.

So as not to be accused of making an incorrect interpretation of ALK's marketing, I will post exactly what their website says about this feature. Notice the phrase "Real Signposts." When I read that advertisement I have an expectation, like in Google Maps and solutions built from HERE maps, that I will be shown real signpost content. In over 7200 miles of driving with CoPilot (2010 trip and this trip), CoPilot seldom showed real signpost content.

In my experience, CoPilot's database does not have knowledge of what the real signpost physically has written on it. Instead, I believe that ALK is simply using knowledge of the crossing or connecting street to generate what it thinks the signpost might say. In some cases that really is what the signpost says on it. In those cases, CoPilot happens to get it right, by chance. In many other cases, the signpost does not contain the crossing or connecting street. Instead, it contains city names, attractions, local road names (instead of highway numbers) and other subtleties. In 7200 miles, I've observed no case where CoPilot accurately shows the real signpost content when the real signpost contains something other than the crossing or connecting street.

Here is just one example of a real signpost image from CoPilot in Kansas City. This sign in real life says "Des Moines." On a long trip like this, literally dozens of examples could be provided like this.


I'll end this particular issue by simply calling it creative marketing though my personal feelings about ALK's advertising of this particular feature are stronger than that.

Speed Limits - Why Bother?
CoPilot RV displays speed limits and also has a nice feature that provides speeding warnings at a user-defined threshold above the speed limit. Great feature. However, that assumes that the speed limits are actually accurate. In my experience, that isn't the case.

The speed limits errors are symptomatic of a larger problem which plagued my experience with CoPilot in 2010 and remains an issue on this trip - map quality.

I found speed limit mistakes on large sections of freeways in almost every state that we traveled on this trip. I'm not talking about missing the change in a speed limit by a mile or so. I'm talking about entire freeways having the wrong speed limit when that speed limit has been set for years.  I'm not talking about wrong speed limits in towns on highways (this could be hard to track because individual towns change them) because I've not found CoPilot to even include speed limits on anything other than freeways.

Illinois: I-74 has the wrong speed limit. It changed in January 2014.
Kansas: I-70 has the wrong speed limit. It changed in the summer of 2011.

In addition, I found incorrect freeway speed limits in Missouri, Oklahoma and Texas.

Maps
In my 2010 blog post, map quality was my biggest disappointment and complaint with CoPilot. Sadly, while the route/ETA calculation has been fixed with the ActiveRoutes functionality, I find no material change that improves the map situation in CoPilot.

CoPilot does update their maps. A completely new freeway interchange was recently added in my home town. I submitted it through CoPilot's MapSure program and it was added in the next quarterly map update. That being said, the problem appears to be what is not reported through MapSure. So, while CoPilot's maps contain this new interchange that I reported, they still have an old freeway configuration in another location in my town (I did not report it) that has been complete for 2.5 years. This is a major US freeway. If major US freeways are outdated by 2 or 3 years, what is the condition of CoPilot's map data on city streets and residential subdivisions?

I have no idea what ALK's internal map update process is. What I do know from experience with the app, is that it isn't good enough. I have raised the map quality issue with ALK in a few ways including their support forum and social media. In all cases, their only response has been to point me to the instructions for submitting map updates to them. On their support forum, they pointed out the link to MapSure then took the unusual step of locking the forum from further comment.

There is much enthusiasm in the GIS industry about user-contributions to maps and crowdsourcing. However, two important distinctions must be made here. First, when it comes to Google MapMaker, Waze and OpenStreetMap, user contributions are used to make really good maps even better. In the case of ALK's maps it is my opinion that they are not starting with "really good maps." Second, customers are paying for this product. Sure, other paid map providers have similar programs. HERE has Map Creator and TomTom has Map Share. However, it is my experience with HERE and TomTom that even without user submission, within a few quarterly updates the internal processes at HERE and TomTom discover and implement many map changes. I readily admit that even with 7200 miles of travel with CoPilot, I have personally experienced a small amount of the ALK map database. However, even with that limited experience, there are enough map errors involving roads that have not changed or have been complete for 2, 3 or more years to question ALK's map update process, especially when directly compared to competitor's maps.

Some examples....On this trip I experienced incorrect freeway exit numbers on I-44 in the middle of St Louis. These were changed in the spring of 2014 and have yet to be updated by ALK. You'll notice in my 2010 blog post that I experienced incorrect freeway exit numbers on that trip as well, on the Florida turnpike.

I encountered problems with the app suggesting maneuvers that are impossible to make in a RV and should not be suggested by any RV navigation app.

Here is an example of CoPilot routing me to a gas station that is located on a major divided highway by routing me instead, in a RV, while using the app's RV mode, onto small residential streets and suggesting that I access the gas station by using a dirt cow path.


I'm not sure if a U-turn was legal at this major intersection in Albuquerque, NM but a RV navigation app, set in RV mode, should never, in any circumstance, suggest this maneuver. Ever.


I also encountered several instances of routing busts. This included:
  • Two instances of the routing not using the separate right hand turn lanes (they were correctly shown in the app) and instead showing a 90-degree right hand turn from the go-forward lanes of the + intersection.
  • One instance of the app suggesting that I travel the wrong way into oncoming traffic of a divided highway, then turn right across my current lane in order to make a right turn.
All three of these errors were located on rural state highways in Oklahoma and Texas. Evidence by the condition of those particular roads, they haven't been changed in many, many years.

There were two instances of this inexplicable maneuver that should be a simple right turn:

 

Conclusion
CoPilot RV is a feature-rich app that easily replaces a dedicated navigation device and matches it feature-for-feature. It's advanced route planning features and ability to build, save and recall multiple routes, each containing multiple stops, is unequaled in the smartphone navigation space and extremely attractive to RV drivers.

The app's route selection is very good, picking routes that I subjectively, would pick myself. The resulting ETA is very good assuming ActiveRoutes is purchased and activated.

The knowledge (and visual confirmation within the app) of RV road restrictions, low clearances and other RV-specific attributes provides peace of mind to the RV driver.

All of those positive qualities are undermined by the user experiencing outdated maps, incorrect speed limits and inexplicable routing anomalies presumably caused by old map data. I remain wholly unconvinced that map updates via CoPilot user submissions is a viable model for maintaining a quality product going forward. As more and more users see the experience of having Google Map Maker edits pushed to Maps within hours, it is my opinion that users will not tolerate road changes not being dealt with by ALK for years. It remains unclear if ALK will get the message and change. A quick read of CoPilot reviews in the iOS and Google app stores will show that map quality is a common CoPilot complaint and has been for a long time. There is no outward evidence based on this user's experience, that ALK has acted on that customer criticism.

An app with CoPilot's feature set, CoPilot's reliability, OpenStreetMaps' timeliness, ActiveTraffic's coverage and ActiveRoutes ETA impact is a winner. Fix your maps ALK. Stop undermining an otherwise great app.