Mapping Uncharted Territory: Developing Indoor Positioning Tools
Good news for those of us without a sense of direction: CEE’s Saurabh Taneja has developed a tool to improve the performance of map-based navigation system for the indoors. Taneja recently graduated with a PhD in Advanced Infrastructure Systems, during which he was advised by Professor Burcu Akinci, Adjunct Professor Lucio Soibelman, and Dean of the College of Engineering and Thomas Lord Professor of Civil and Environmental Engineering Jim Garrett. In his thesis, Taneja outlined a framework that he developed for improving the accuracy of indoor positioning via a technique known as map-matching.
Taneja’s research addresses the increasing need for a reliable indoor positioning tool, which is tied to a rise in the complexity of indoor environments such as airports and office buildings. In situations such as emergency response and disaster management, such a tool would be invaluable in guiding building occupants to safety. It could also be used to provide support to construction and maintenance workers or to help individuals with disabilities navigate unfamiliar environments. However, satellite-based technologies such as GPS won’t do the job; the signals from the satellites cannot penetrate buildings and walls. So what’s the alternative?
Cue an innovative technique called map-matching, which was originally developed to improve the accuracy of outdoors GPS. Map matching is overlaying raw positioning data—for instance, data obtained from wi-fi or GPS—onto the map of a physical environment, then correcting the position of the sensing source to accurately place you in the map. “We need map-matching because raw positioning data does not always reflect the limitations on your motion,” Taneja explained. “For example, if you overlay raw positioning data on a map, it might suggest that you’re walking through walls or flying in space; map matching allows you to say, no, you did not walk though this wall, you went through this particular door.”
Taneja saw the potential for improving the accuracy of positioning data in indoor environments through map-matching. While map-matching for GPS and the outdoors is well understood, much less is known about achieving it indoors. “Outdoors, GPS provides your position in latitude and longitude, and then it is overlaid on a map and converted to a street address,” Taneja said. “But inside buildings, you can describe your position as a room number, a floor, a zone… not just as a point. There are different formats of indoor positioning data, and to map-match those different formats, you need different types of maps.”
Taneja developed a series of algorithms that generate maps of indoor environments using digital building information models and then map-match raw positioning data onto them. To ensure that his approach would work for a variety of buildings, he designed his algorithms to automatically generate three different types of maps for six types of indoor environments that vary in shape and density of spaces.
Taneja envisions his work being used in a mobile platform such as Android. An application such as Google maps could use the tool he developed to decide which type of map to use for map-matching in a particular building. For instance, imagine a traveler is navigating an unfamiliar airport to catch a connecting flight. An app on the traveler’s smartphone would select the type of map for map-matching that best represents that airport’s layout. The traveler would then receive an accurate estimate of their position and could take the quickest path to their destination.
Though he worked on three different topics during graduate studies, Taneja was most drawn to this one because of his long-standing interest in maps and navigation. “Even when I was a very small child and my father used to work in New Delhi twenty miles from our home, I had memorized the way to his office, and I still remember it,” he recalled. “I would always keep an eye on the road and ask myself, where are we turning, is this the shortest way?”
This summer, Taneja has joined a Pittsburgh firm specializing in supply chain management to work as an algorithm developer. As for what the future holds, he’s dreaming of data. “I have always wanted to work with the kind of huge, enterprise-level datasets that we see today, such as data coming from trade markets, news streams, and live video feeds,” he said. “I’d like to get some more experience in that field, and will then decide whether to return to research or stay in industry.”