AR Browsers Head To Head – Test 2
Whereas part 1 of the head to head challenge focused on finding food, part 2 of the augmented reality browser head to head challenge focuses on using the browsers to find more information about your surroundings.
The scenario is you are travelling and you see something of interest, can an augmented reality browser tell you what it is?
About the test
The test took place in London along the River Thames. I tested 3 augmented reality applications:
- Wikitude’s Qype and Wikipedia search
- Layar’s Wikipedia and the Unlike Global Guide layer from Unlike Media Limited
- Robotvision’s default Wikipedia search
With each test I switched to the relevant map mode to double check my results.
The test begins at London bridge, looking along the river I can see HMS Belfast (an ex-Royal Navy Town-class cruiser) and just behind it Tower Bridge.
Layar gets off to a bad start as neither of the layers find any of the two reference landmarks, switching to the map view confirms that neither of the landmarks are found. Using the Wikitude Wikipedia search we draw a blank on both the landmarks, but the Wikitude Qype search does find HMS Belfast. Robotvision uses the Wikipedia database and it fails to find any of the landmarks.
HMS Belfast is a well known London landmark visited by thousands of tourists each year and Tower Bridge is a symbol of London so I was really surprised that it wasn’t found. The browsers had no problem with the other bridges along the Thames even the insignificant ones.
As I worked my way down the Thames I came across the London Dungeons so thought I would stop and do a quick test. Both the Layar layers along with the Wikitude Wikipedia search found the dungeon. But both Wikitude Qype and Robotvision missed the landmark.
Not far past the dungeons is the Golden Hinde, a ship captained by Sir Frances Drake. It wasn’t going to be part of my test but since I was standing in front of it, just like a tourist I thought I would take the chance to find out more. The problem here is there is not a lot of space so you are literally on top of the ship. GPS is accurate to something like 50 meters so being so close to the subject actually didn’t help. My initial tests showed that none of the browsers found the Golden Hinde but to be sure I switched to map mode and looked at the POIs on the map.
Once in map mode only Layar and the Wikipedia layer found the object. In the photo you’ll see that the accuracy Layar is reporting is 1.5km (about a mile to the rest of us), so the marker appears on the extreme left, almost with the ship out of view. I put this down to the fact that it’s a really closed in area with lots of buildings.
There are several other tourist attractions here, Southwark Cathedral (the oldest cathedral in London) and Clink Prison which dates back to the 12th century. Only Robotvision and Layar (Wikipedia) found these landmarks.
A short walk from the Golden Hinde you can see St Paul’s Cathedral on the other side of the Thames. Layar and both the Wikipedia and Unlike layers failed to find the cathedral (confirmed via the map view) as did Wikitude (Wikipedia) but Wikitude Qype and Robotvision had no problem finding the landmark.
The London Eye is probably London best know tourist attraction, so it was a good sign that all the browsers had no problem finding the landmark. What helped here was I was on the other side of the river with a good distance between me and the target so GPS accuracy wasn’t so much of an issue. Actually this was one of the best tests because it really was a case of point the camera and the visual marker would come into view very near the object.
The last test was the Houses of Parliament. Layar (Wikipedia) had no problem with finding the target as did Robotvision. Wikitude (Wikipedia) failed to find the target, but after refusing to give up with Wikitude (Qype) I did manage to find it. The location is slightly wrong so it didn’t appear when I pointed in the correct location. The only interesting thing Unlike found was Indigo02 and I have no idea what that is.
Before we move on to the results here are some of my experiences with the various apps.
Both Layar and Wikitude drove me crazy with having to work in Kilometres, I know I mentioned it before but it is really annoying not to be able to switch into a localised unit of measurement particularly when trying to judge objects over large distances.
Every time you open Wikitude you have to reset the distance or it defaults to a range of 67.69 km for Wikipedia and 11.41 km for Qype. Wikitude also crashed every time I changed the POI database so I would have to restart the app and then reset the distance. That got annoying pretty quickly with all the tests I was doing.
I like the fact that in Wikitude you can select multiple databases, in real life you are probably going to have your defaults selected so will be less likely to fiddle like I did. With Layar you can only use one POI at a time and given the differences between the Unlike and Wikipedia databases you are never really sure if you are missing anything.
Robotvision was kind of a surprise, it used miles which was good but again every time you opened the app the distance reset to 3 miles (I guess the distance is somewhat reasonable). Robotvision was easy to use as there was no switching between databases, you have the choice of Twitter, Bing, Flickr and Wikipedia. Robotvision is actually a pretty good augmented reality application and deserves far more recognition that it receives. It easily competes with the big boys so well done to the developers.
What’s interesting however is the differences between the Wikipedia searches. For Layar the database that performs the best is the Wikipedia search, for Wikitude it’s the Qype search. I think it’s an interesting difference and deserves further investigation.
The scores then, in this head to head challenge each browser was asked to find 7 well know London landmarks so we’ll award 1 point for each landmark found.
- 1st place Wikitude with 5 points
- 2nd place Layar with 4 points
- 3rd place Robotvision 3 points
Can you use an augmented reality browser to find out information about your surroundings? The answer is yes.
A couple of tips:
Generally it’s a lot easier to have some distance between you and the subject to combat the GPS accuracy. If you are too close you may have to pan around to find the object. Secondly it’s sometimes easier to find the object on the map view first, that way when you switch the camera view it’s selected and saves you searching around.