Tag: Zeitspace

David Trueman, of CycleWR, wrote at the Zeitspace blog recently about OpenStreetMap (OSM) and why we use its data in Cycling Guide. One of the reasons is that we can update OSM as new cycling infrastructure opens in the real world or as we discover issues in its data. Here’s how data-related improvements to our routing happen.

A big part of our testing early on was riding our bikes and experiencing Cycling Guide routes directly ourselves. That doesn’t scale, though, and we invited other people to get early access to Cycling Guide in order to test it and provide feedback. That testing has been a great success, especially in revealing issues in our routes. Here’s a recent example of feedback from one of our early access testers: 

King St between Marshall and Regina is finished construction and has above grade bike lanes that are pretty good although you have to be cautious of pedestrians and parked vehicles. Routes are avoiding that section and taking inferior streets .

I’d been watching King Street North in Waterloo over the summer wondering when the opening would happen, and hadn’t realized that it had. Neither had David. He checked OSM and saw that there was nothing in its data yet about the new north-bound and south-boud cycleways. It was time for a quick reconnaissance trip. I cycled over to King Street South and Willis Way, and from there headed north on the dedicated cycleway. And as reported by our Cycling Guide tester, the new section is now open and I was able to ride all the way to University Avenue. I was able to ride back as well on the southbound cycleway, with the exception of a short segment between University and Bricker.

Having confirmed the physical state of this bit of the world, I let David know. He made an update in OSM, which is now awaiting review by other OSM contributors. We’ll see those cycleways on Cycling Guide routes and maps soon.

But updates to OSM aren’t the only way that we improve our cycling routes. We’ve had multiple testers let us know about rough trails that aren’t really suitable for many cyclists. They’re not appropriate for cargo bikes, or bikes pulling trailers, or simply just for less-experienced or less-confident cyclists. Here’s one report:

Route did not suggest going directly down Doon South but rather to take a trail just after Winding Wood Cres. The trail had an intense, steep hill that would be challenging for most cyclists.

To address this recurring issue, we’ve been updating the way that we process OSM data for use in our routing. As a result, we’re avoiding more and more of these kinds of rough trails. As I’ve mentioned previously, reports of real-world experiences from our group of testers is a big part of making sure that Cycling Guide routes are as good as they can be.

Here’s part of another recent example of a routing challenge that was quite tricky to diagnose, but simple to resolve:

I’m surprised that it didn’t suggest using the bike paths north of University Ave (near MacGregor Cres.) since that route takes about half the time and has less interaction with cars. I’m wondering if this route was intentionally avoided or if those paths just aren’t included in the map. 

The routes that Cycling Guide provided in this case were far more indirect than they should have been, and it wasn’t at all clear why. This one was trickier to track down, but it eventually turned out that in the OSM data one short segment of the Laurel/Forwell Trails didn’t explicitly allow bikes, while all the other segments do. That prevented Cycling Guide from using those parts of the trails. This feedback resulted in another quick OSM update by David (again, pending review by other OSM contributors).

Sometimes we get reports of problems that are already close to being addressed. Here’s some recent feedback we received from another of our early access testers:

The road from Wellington at Walter to Park St, that runs between the GRH helipad and the badminton club, is a private road with gates. A route tried to take me through there today, but it is not passable.  

I checked, and the issue was that we weren’t handling that private road correctly. That is, the OSM data is correct, but the way we were processing it needed improvement. The good news is that I checked with the newest internal routing data, and confirmed that we already have a fix that excludes private roads like this one from routing. That fix will be out in the world soon.

Engaging with people who joined our early access to Cycling Guide group has been, and continues to be, a great success. Sometimes the problems people have reported have fixes that apply across the Region because of improvements in how we process OSM data. Sometimes the fixes are to OSM data itself. In all cases, though, the result is better cycling routes for everyone.

If you’re interested in getting early access to Cycling Guide yourself, sign up right now at CyclingGuide.app. We’d love to hear from you.

This post originally appeared on the Zeitspace blog.

The Cycling Guide mobile app, which Zeitspace is creating with CycleWR, has reached a milestone. After an extended period of internal testing, we’ve put it in the hands of several people who are experienced and confident cyclists to see how well it works for them. But the app is actually aimed at people who aren’t experienced and confident cyclists. Why have we gone with these experienced and confident cyclists for early testing? It’s because we’re particularly interested in learning about the quality of the routes that Cycling Guide provides, and we want to start with people whose cycling experience and knowledge of cycling in Waterloo Region will better equip them to assess those routes.

Here’s a little bit of an overview of the routes. One of the things that we want Cycling Guide to do is provide low stress routes for people who otherwise may not feel confident or comfortable enough to ride in Waterloo Region. The routes that Cycling Guide presents on the map show cyclists what they can expect in terms of stressful (or not) car traffic when cycling along a route.  When creating a route, the app classifies streets and paths into four groups:

• Good for anyone
  You’ll ride with few or no cars
• Good for most cyclists
  You’ll ride in lighter car traffic
• Good for confident cyclists
  You’ll ride closer to cars
• Good for strong cyclists
  You’ll ride in heaver car traffic

The resulting Cycling Guide routes are often low stress, using protected cycling lanes, shared paths, and other car-free options. That is, Cycling Guide routes will typically favour the lowest stress paths and streets, the ones that are good for anyone. That’s great for less experienced cyclists.

Of course, this is what we intended. How close have we gotten to achieving this? So far the feedback is great. Here’s what some of our testers are saying:

• “Passed through a park with construction on the trail which was not noted on the map. It was passable but looked like it had recently been closed.”
• “Route did not suggest going directly down Doon South but rather to take a trail just after Winding Wood Cres. The trail had an intense, steep hill that would be challenging for most cyclists. Doon South at least is flat. Then lost the trail/route through a GVTA trail that was not appropriate for cyclists. Felt out of the way and not very direct.”
• “I hit a dead end near a pond looking for this route, and couldn’t see it. I ended up getting back on Laurel Trail, to land near the top of this mystery segment, so I got back on track.”

And finally:

• “I just want to emphasize I really love the problem this app is looking to solve.”

As the first three examples show, we have plenty of room to improve. But it’s great feedback as it all addresses exactly what we’re looking to learn about in our current testing phase: the quality of Cycling Guide routes. I also like the fourth piece of feedback, as it confirms that we’re going in the right direction!

Our work continues, and Cycling Guide improves every week.

If you’re interested in getting early access to Cycling Guide yourself, sign up right now at CyclingGuide.app. We’d love to hear from you.

This post originally appeared on the Zeitspace blog.

I want to provide an update on Cycling Guide, the mobile app that Zeitspace is working with CycleWR to create. One of the fun aspects of working on Cycling Guide has been something that software product teams don’t always enjoy: testing

More specifically, as the team has worked on the app we’ve needed to assess the quality of the cycling routes that it provides. As a bit of background, we’re building Cycling Guide on a foundation of geospatial data from OpenStreetMap. It’s a fantastically detailed and accurate resource. We’ve been able to process the underlying OSM data that we use to better understand and visualize our routes, which has been a great starting point for our routing approach. But we’re also doing exploratory testing of our cycling routes, something familiar to software product teams.

In our case, it started by looking at the routes on screen. As Cycling Guide is  focussed on Waterloo Region, we’re assessing routes on streets that we know at least a little. As work progressed we could easily identify anomalies in the routes we looked at and address them in development. We also defined some specific trips that we could review again and again, looking to see how updates to data or our routing engine affected the routes. For example, there’s a two-way protected cycling lane along the otherwise one-way Erb St between Peppler and Caroline in Waterloo that we’d expect to see used for some routes in either direction. And we have a few test trips where we would reasonably expect that routes would use the Iron Horse Trail, Spur Line Trail, or Laurel Trail. My own favourite test trips are to local breweries. 

In addition to software updates and data updates, being able to tweak routing settings while testing and seeing the results on a familiar street and path network has made the work a lot of fun. We’ve also compared our routes to what some other mobile mapping apps provide, and the results are encouraging. In many cases we prefer the routes that Cycling Guide provides.

Exploratory testing on screen was a place to start and helped the team make great progress. Things got more fun when we started doing exploratory testing while on bikes in the real world. The earliest field testing that our team did was pretty simple, as the app was extremely limited in functionality. For example, riding along Iron Horse Trail and periodically checking the current location marker on a map wasn’t really delivering much value, but the functionality worked. As Cycling Guide grew in capability I started trying more ambitious rides myself. Two recent cycling trips provided unexpectedly strong validation that the app is already delivering value.

The first was a ride with my son from Uptown Waterloo to St. Jacobs and back. The route took us on paths that we didn’t know about, or hadn’t been able to find previously. And the app was great to use while riding even in a still-limited form. The second was a ride I did from Uptown Waterloo to Cambridge and back. Again, Cycling Guide took me on paths that were new to me. Both rides were mostly on protected lanes or shared paths that were clear of cars. And both rides were quite scenic in places. The rides were unexpectedly exhilarating experiences. My typical rides are either running errands or longer recreational rides. Having a way to expand my cycling repertoire will provide great value for me.

As I like to cycle anyway, this exploratory testing phase has been great fun. In addition to testing our app, it has me exploring Waterloo Region by bicycle and making new discoveries. Even at this early stage we’ve been able to make some small improvements in OpenStreetMaps data as well. So it’s been a win on multiple levels.  Sometime soon we’ll be expanding our testing to include some outside testers who aren’t a part of the project team. We’re excited to see the results.

This post originally appeared on the Zeitspace blog.