SUBSCRIBE TO OUR BLOG
So when it comes to the future of routing advice, it’s perhaps no surprise that there’s a team at HERE working to make future products more personal and personalisable.
Daniel Rolf – we talked to him last about the science of traffic – now heads up our Platform Services Innovation office – one of our research teams devoted to problems like routing.
“Historically, calculating routes has been a cartographic problem," Daniel tells us. "We have an enormously detailed map. So routing becomes a question of plotting your location and destination into that and finding the path with the best time.”
There’s always research to be done, though, and ways we might improve the experience.
“Two years ago, we ran a project that abandoned maps entirely – not a very HERE thing to do!” says Daniel. “And investigated what would happen if you just used probe data – the signals from GPS-enabled devices.
“One of our researchers uploaded all his personal bike trails and times. So that way, we could provide an accurate ETA for his next journey to work that’s based on his actual cycling speed and the shortcuts and approaches he uses, not an abstract computer model of an average cyclist. And we can potentially make routes for places with no maps.”
This use of personalised probe data to influence routing set a second research project in motion.
“Currently most GPS software will give a choice of more or less three routes, and the ETA that’s expected for each of them.”
“So we asked, can we add a more social aspect? What if we showed the percentage of people who would use each choice?”
If three routes are shown, as normal, but you see that 70 per cent of people take a route that’s not the fastest, then you can be pretty confident there’s a reason for that. Maybe the fastest route is also full of potholes, or maybe it’s always very busy with traffic? Also, knowing what people usually do might help you to make a decision when all three routes have the same ETA and all look equally nice on the map.
“People like to see information, and love to know what other people are up to. It actually went down very well as a trial project with users.”
“The next thing we looked at was using a combination of probe data from the individual alongside our normal routing algorithm.
“When people are close to home, they tend to have their own routes. And when a GPS system tells them to go a different way, then they’ll either think it’s wrong or ignore it anyway.
“So maybe we can do something with that, and use your historical probe data closer to home – so it follows your established preferences, but mixes in other people's probe data when you’re travelling further away.
“We might know those routes are suboptimal – but they feel much more human.”
“We then take that to the next advance, and start to build profiles. We can look at the way you drive, and when you go somewhere, we can show how people like you would approach this journey.
“A good application might be getting around a foreign city you’re visiting. It would feel like having a local soulmate that exactly imitates your driving style, e.g. it avoids left turns because it knows that you hate left turns.
“We might make those profiles selectable.” – you might normally use the ‘get me there quick’ profile, but then sometimes switch to ‘I’m stressed today: take me a route that’s less busy’ or perhaps ‘get me there quick but avoid rainfall.’
The exact shape and timeline of future product releases is, of course, confidential. But it’s always exciting to learn about the possibilities.
Which of these research ideas is most exciting to you?