Analysis of Google’s Bike-There Feature: Part I

thumb.jpgPhoto: Chloe Forsman

It’s been nearly a month since Google made their game-changing announcement about the addition of a “Bike There” option to Google Maps.  Since that time, there have been numerous reviews and assessments about the utility of the new application – some positive, but also many that point out the limitations of Google’s new tool.  Many people have noticed that Google sends them on roundabout routes or through inaccessible places, while others have had more success.  Therefore, you might be wondering at this point whether or not Google is the best resource available for figuring out what route to take in order to get from by A to point Z (and beyond!) by bike.  In this series, we are going to try to help answer that question and attempt to determine if Google is the best available resource for bike mapping and route-finding at the moment, what are its advantages and disadvantages, and what other resources are available for finding a good route by bike.  We will begin by reviewing the Google Bike-There feature and the main principles that Google’s engineers used to build the algorithm that spits out the directions.  In following posts, we will review other bike mapping tools, which we will compare and contrast to Google’s Bike-There feature, and try to determine the best method for mapping bike routes.  Our goal for this series is to provide some insight into the challenges and issues of bike-route mapping, and we hope that you join the discussion in the comments section.

Behind the Scenes — How Google’s Bike Maps Work

The push to get Google to incorporate bike directions into Google Maps has been going strong for quite some time now, but Google reports that adding such directions presented quite the engineering challenge.  Google uses a few key features to develop the algorithm that generates a bike route, using the already-existing network of streets in their mapping system, which are summarized below.

  1. Bike Trails – These show in dark green when you generate a Bike-There map.  Google worked directly with the Rails-to-Trails Conservancy to find as many trails to incorporate as possible.  The algorithm is weighted to send cyclists on trails as much as possible, as long as it doesn’t send them too far out of the way.
  2. Bike Lanes – Google has information about dedicated bike lanes for 150 U.S. cities, which they used to build into the algorithm.  Bike lanes appear on a Bike-There map in bright green, and they are also weighted as a priority in the algorithm.
  3. Recommended Routes – These are routes from cities that have information about other good roads for cycling, which may not have an official bike lane.  These roads are indicated by a dashed green line in the Bike-There map.
  4. Uphill Slopes – In order to avoid hills (because, according to Google, nobody likes riding up hills… though I beg to differ…), Google developed a model that takes into account power (exerted by the cyclist), the slope of the road, wind-resistance, and speed.  If the model shows that a given route requires an inordinate amount of exertion (aka too much power required) and will be too slow for time efficiency, Google will send you on an alternate route that avoids the climb.  I could not find out what Google defines as “too slow” or “unreasonable degree of exertion”.
  5. Downhill slopes – The model will also help cyclists avoid roads with too much downhill or descending, which can be tiring  or disconcerting due to the unnecessary amount of braking required.
  6. Busy roads – In order to keep cyclists off busy roads, the algorithm basically uses the inverse of the Drive-There algorithm in order to avoid arterials and freeways.
  7. Intersections – Lastly, the algorithm avoids busy intersections with heavy traffic (car) and long waits.

If you want to read this information directly from the horse’s mouth – click here.  It is important to note that this system is still in beta, and Google is continuing to add and update their maps using feedback from users.  So if you found an issue, make sure to notify them.

Picture 2

Review of Google’s Bike-There Features

I must admit, I have a few opinions on the underlying principles that Google used to develop their algorithm.  However, I am certainly not an expert in building such algorithms, and I realize the inherent limitations and difficulties of undertaking such an engineering project.  Nonetheless, there are some underlying biases and (possible) misconceptions about cycling built into the algorithm, which make the Bike-There feature unappealing to me in some ways.

Picture 1Google Maps Bike Routes around University of AZ

First, I am curious about Google’s decision to avoid major intersections.  When I experimented with numerous routes that I use in Tucson, I discovered that Google’s directions, do indeed, avoid major intersections.  However, in doing so, the directions take a route that requires crossing major roads without the benefit of a traffic signal found at an intersection.  Sure, the directions managed to avoid the major intersections, but it’s not always the best riding habit to avoid such areas, which, at least in Tucson, often have infrastructure in place for cyclists to trigger the traffic lights and cross safely when the light is green instead of darting across four, fast lanes of traffic.  How do you feel about the avoidance of major intersections where you live?

Second, I have a little trouble with the prioritization of bike trails over established bike lanes (and/or recommended routes).  According to Elaine Filadelfo, a Google Maps spokeswoman:

It’s a pretty complex algorithm that looks at all of the variables [including hills, traffic, road type and many others] and tries to come up with the most efficient final route. For example, it will prioritise putting you on a trail or a road with a bike lane and weigh that against how far out of your way it might send you or how hilly that terrain might be. We look at all of the variables in conjunction, but I would say the strongest factor we look for is to put you on a bike trail. (via Bike Radar)

At least in Tucson, the official bike trails are often quite congested with pedestrian traffic (usually of the recreational nature), which can make commuting somewhat more time consuming, as one has to ride slower.  Given that a city like Tucson covers a large area (as do many U.S. cities), efficiency in bicycling can be quite important, so prioritizing bike trails over bike lanes can be somewhat problematic, especially in circumstances where bike trails are congested or out of the way.

Finally, I would love to know more about Google’s model to avoid uphill slopes.  If anyone knows of any additional information, I’d appreciate your insight.

In general, and despite my criticisms, I think Google is off to a great start.  In fact, it’s quite incredible that Google has finally incorporated bicycle directions, and I am looking forward to the impending improvements.  I completely agree with Peter at GoogleMapsBikeThere, that because almighty Google has acknowledged the importance of bicycling directions, more and more influential groups and people are going to take note.  So, thanks, Google!

Next up, we will review other bike route mapping tools, as a matter of practicality.  What’s the best way to get from point A to point Z by bike?