A few nights ago, Christina and I were driving down to Corsicana to pick up our niece for the weekend. While driving, I decided that I needed to find a QuikTrip to get a good drink. Having just installed Google’s new Voice Search update, I tapped the shortcut I’d setup and said ‘Find the nearest QuikTrip‘. (Normally I’d call it a QT, but I figured I’d help the software out a bit). As expected, the phone responded a few seconds later with a list of QuikTrip gas station locations that were ‘near’ us. Good so far. Unfortunately, the two that it found were both in the OPPOSITE direction that I was currently heading down the freeway at 70MPH.
I’ve struggled with this before, and I suspect I’ll struggle with it again – while the latest and greatest location-based services are helpful, they all miss on a key component, and Nokia is the only company that I have heard use this word in the same sentence as location: context. In the situation above, my contextual location would have told the software that since I was heading down the freeway at 70MPH (and had been, with maps running, for a few minutes), that I probably didn’t want to turn around and go 10 miles in the opposite direction. My direction and velocity are the context, which should have told the software to look for QuikTrip locations along the highway that I was traveling on, in the direction in which I was traveling.
Another time this problem appears is in the automatic re-routing feature on most modern GPS systems. If I miss my turn (usually on purpose, for testing, but still), all it says for the next 10 minutes is a series of U-turn opportunities. What if I don’t want to make a U-turn, for whatever reason? Why not look ahead in the journey and see if you can re-route me in a forward motion?
It’s this intelligence that I’m anxiously looking forward to from today’s leading GPS applications (those being, in my opinion, Google Maps Navigation, Telenav, and Ovi Maps). Such an intelligence addition might bring the likes of TomTom and Garmin back to relevance in the marketplace, too. Basically, even when not actively navigating, I want to abolish this whole ‘search within an X mile radius‘ thinking from today’s location-based services, and get into more contextual results that take into account other aspects, such as speed and direction of travel.
Have you experienced this, too? What other contextual location data can you think of that would be helpful?
Ricky Cadden 
This where where the compass “should” come into play and I assume the the magnetometer API’s aren’t being implemented correctly.
This where where the compass “should” come into play and I assume the the magnetometer API’s aren’t being implemented correctly.
If it’s telling you to make a U-Turn, that’s because the algorithm used and their heuristics figure that it’s quicker than continuing straight on. Short of highly situational shortcuts and traffic info unavailable to the GPS, they’ve pretty well reached the point of being better navigators than most humans
If it’s telling you to make a U-Turn, that’s because the algorithm used and their heuristics figure that it’s quicker than continuing straight on. Short of highly situational shortcuts and traffic info unavailable to the GPS, they’ve pretty well reached the point of being better navigators than most humans
I think the real problem here is one of data. Think about how much extra data that would add to have contextual relevance to anything and everything you could possibly ask of it? If you ask “where’s the nearest Quik Trip”, the map has: one set of data (a geolocated list of Quik Trip’s) to compare against another set of data (your current geolocation). But if you add context, then you add a third set of data (your relative path of travel through geolocation space), and a fourth set of data (call it the ‘real world obstacles’ set of data, since the s/w would have to know whether something that was 10 miles in the right direction, but on the other side of the freeway, was accessible to you).
For that matter, to refine context even more requires even more data: is the Quik Trip open (hours of operation)? etc etc
I think you are right that context is EXTREMELY relevant to geolocation; but I don’t think that we are quite to the point where a mobile device can handle every set of data in an efficient enough manner to make it work. We’re getting there, but not quite yet.
Then there’s one last thing to consider: datasets are EXPENSIVE. My company licenses lat/long data to the zip+4 level for our software (meaning that it’s accurate to the street level). The cost to us to license this runs into the tens of thousands per year. When you get down to looking at the cost of licensing data to the rooftop level it’s 100’s of thousands. When you get down to full on to-within-a-meter data, you can see why companies like Nokia and Google ust outright buy other companies: it’s cheaper in the long run to buy them then license their datasets.
Imagine, then, the cost of building in 4, 5, 6 other datasets? You better have proven tech that will sell like hotcakes, or it’s just not worth it.
I think that Garmin or TomTom or someone would have gotten their eventually, since that was their niche, but at this point they are so superseded by the availability of smartphones that I don’t see them having the money to do so.
I think the real problem here is one of data. Think about how much extra data that would add to have contextual relevance to anything and everything you could possibly ask of it? If you ask “where’s the nearest Quik Trip”, the map has: one set of data (a geolocated list of Quik Trip’s) to compare against another set of data (your current geolocation). But if you add context, then you add a third set of data (your relative path of travel through geolocation space), and a fourth set of data (call it the ‘real world obstacles’ set of data, since the s/w would have to know whether something that was 10 miles in the right direction, but on the other side of the freeway, was accessible to you).
For that matter, to refine context even more requires even more data: is the Quik Trip open (hours of operation)? etc etc
I think you are right that context is EXTREMELY relevant to geolocation; but I don’t think that we are quite to the point where a mobile device can handle every set of data in an efficient enough manner to make it work. We’re getting there, but not quite yet.
Then there’s one last thing to consider: datasets are EXPENSIVE. My company licenses lat/long data to the zip+4 level for our software (meaning that it’s accurate to the street level). The cost to us to license this runs into the tens of thousands per year. When you get down to looking at the cost of licensing data to the rooftop level it’s 100’s of thousands. When you get down to full on to-within-a-meter data, you can see why companies like Nokia and Google ust outright buy other companies: it’s cheaper in the long run to buy them then license their datasets.
Imagine, then, the cost of building in 4, 5, 6 other datasets? You better have proven tech that will sell like hotcakes, or it’s just not worth it.
I think that Garmin or TomTom or someone would have gotten their eventually, since that was their niche, but at this point they are so superseded by the availability of smartphones that I don’t see them having the money to do so.
That is why you can still not be 100% reliable on navigation softwares yet.
I especially like the case when such software suggest taking U-turns when you can’t do it at all. Other times I don’t because I know that I can take a different route leading me to the same destination, Whereas the navigation software throws me into a longer distance when suggesting a new route.
Long story short I was thrown into a 6 km walk after taking a wrong bus home after work. (I could have cut the route straight, but last time I attempted that I had to walk through some threes and climb over a fence).
Both cases was when using OVI Maps and despite it serves me quite well I can’t say I’ll trust any navigation software without thinking for myself.
I COMPLETELY agree. This has been a huge pet peeve for me and I think the first person to come out with some great navigation software that allows you to search within your established route will make a boat load of money. Also, including a gas buddy like trip planner that would allow you to find the cheapest gas on a route you have planned would be awesome.
I COMPLETELY agree. This has been a huge pet peeve for me and I think the first person to come out with some great navigation software that allows you to search within your established route will make a boat load of money. Also, including a gas buddy like trip planner that would allow you to find the cheapest gas on a route you have planned would be awesome.
Tell me about it!! Forget 70mph; what about 3mph and it’s raining? If you live in the city and use GPS, it would be nice to be able to tell the thing not to search further than 1/4 mile. And better yet, not only sense that I am on foot by speed, but sense that it is raining in my location and I just don’t wanna walk more than 100 yards.
Also, since I haven’t gotten my GPS adapter yet, how is it with public transit? It would be nice to be guided to the nearest bus stop based on things like shortest time (tracking buses in real time) or lowest cost, or again, shortest walking time when it’s raining. Far fetched, I know. But some cities already have real time bus/train tracking so maybe…
About the lowest cost. In Seattle it can cost almost $9.00 round trip now to go from the suburbs to the city. This is because you normally have to use two bus companies and they don’t share the cost. It takes more time, and some clever planning on Google but it’s possible to stick to the city’s main bus line. GPS should be able to do this.
Tell me about it!! Forget 70mph; what about 3mph and it’s raining? If you live in the city and use GPS, it would be nice to be able to tell the thing not to search further than 1/4 mile. And better yet, not only sense that I am on foot by speed, but sense that it is raining in my location and I just don’t wanna walk more than 100 yards.
Also, since I haven’t gotten my GPS adapter yet, how is it with public transit? It would be nice to be guided to the nearest bus stop based on things like shortest time (tracking buses in real time) or lowest cost, or again, shortest walking time when it’s raining. Far fetched, I know. But some cities already have real time bus/train tracking so maybe…
About the lowest cost. In Seattle it can cost almost $9.00 round trip now to go from the suburbs to the city. This is because you normally have to use two bus companies and they don’t share the cost. It takes more time, and some clever planning on Google but it’s possible to stick to the city’s main bus line. GPS should be able to do this.