Yahoo! Researchers Have Developed A GPS Algorithm To Find 'Emotionally Pleasant' Routes

Mapping services and GPS devices are generally used for finding the shortest routes between two points while walking around or driving a car. Over the last few years, GPS technology has improved with features like the ability avoid tolls and construction. Now a team of researchers at Yahoo  Labs in Barcelona, Spain want mapping services to offer pedestrians with a way to find the most scenic routes while heading towards their destination.

Daniele Quercia and Luca Maria Aiello of Yahoo Labs partnered with University of Torino's Rossano Schifanella to develop a mapping algorithm that could be used for finding the most "emotionally pleasant" routes. Their findings were published in a paper called The Shortest Path to Happiness: Recommending Beautiful, Quiet, and Happy Routes in the City by Cornell University's arXiv.

Quercia and the research team started by gathering 3.7 million images from various locations in London using Geograph and Google Street View. The images were uploaded to UrbanGems.org, where opinions about the beauty of each location were crowdsourced. UrbanGems visitors voted for the more beautiful place out of two photographs at a time. The more beautiful the location, the more likely it would become part of a suggested route by the algorithm. The researchers considered votes from about 3,300 visitors.

After the beauty scores of locations were determined, the researchers plotted out maps of the beautiful routes. On average the beautiful routes were only 12% longer than the shortest routes, making them viable alternatives to routes that have a focus on tourism or entertainment. "For example, paths recommending touristic attractions tended to be half-a-day touristic experiences (twelve hours), and those capturing people's salient experiences tended to be 60% longer than the shortest paths," said the researchers in the published paper.

The beautiful routes generated by the algorithm were tested by thirty people who live in London and know the area very well. The thirty people assessed the recommended paths and agreed that the routes selected by the algorithm were more beautiful.

Crowdsourcing opinions could become tedious when expanding the algorithm to other cities so the research team decided to automate the process using metadata and tags from Flickr photos. The research team analyzed five million photos of the same locations that were considered more beautiful in the initial experiment to determine indicators of beauty score correlations. Collecting locations of photos that have a higher number of uploaded photos and positive comments ended up producing a list of beautiful places in a city.

To test out the automated process using metadata from Flickr, the research team applied the algorithm to the city of Boston. Fifty-four participants were selected to compare the directions based on the metadata from Flickr with the shortest routes. The participants rated the path generated with the Flickr metadata were "on average, 35% more beautiful than the shortest paths."

Based on feedback from participants, it was determined that there are several limitations with the algorithm. For example, the time of day, day of the week, time of year and weather conditions can affect how beautiful an area is. One participant said that the St. Paul's area of London is nice on the weekends when the city is quiet, but it can be an unpleasant experience during busier times.

The research team plans to build upon this analysis by designing a mobile app and testing it in different cities across the U.S. and Europe.

What are your thoughts about this study? Would you prefer to take scenic routes while on a trip? Let us know in the comments.