科学家研制出室内导航系统

University of Nevada, Reno computer science engineering team Kostas Bekris and Eelke Folmer presented their indoor navigation system for people with visual impairments（缺陷） at two national conferences in the past two weeks. The researchers explained how a combination of human-computer interaction and motion-planning research was used to build a low-cost accessible navigation system, called Navatar, which can run on a standard smartphone. "Existing indoor navigation systems typically require the use of expensive and heavy sensors², or equipping rooms and hallways with radio-frequency tags that can be detected by a handheld reader and which are used to determine the user's location," Bekris, of the College of Engineering's Robotics Research Lab, said. "This has often made the implementation³ of such systems prohibitively expensive, with few systems having been deployed⁴."

Instead, the University of Nevada, Reno navigation system uses digital 2D architectural maps that are already available for many buildings, and uses low-cost sensors, such as accelerometers（加速计） and compasses, that are available in most smartphones, to navigate⁵ users with visual impairments. The system locates and tracks the user inside the building, finding the most suitable path based on the users special needs, and gives step-by-step instructions to the destination.

"Nevertheless, the smartphone's sensors, which are used to calculate how many steps the user has executed and her orientation⁶, tend to pick up false signals," Folmer, who has developed exercise video games for the blind, said. "To synchronize⁷ the location, our system combines probabilistic（概率性的） algorithms and the natural capabilities⁸ of people with visual impairments to detect landmarks¹⁰ in their environment through touch, such as corridor intersections¹¹, doors, stairs and elevators."

Folmer explained that as touch screen devices are challenging to use for users with visual impairments, directions are provided using synthetic¹² speech and users confirm the presence of a landmark⁹ by verbal confirmation¹³ or by pressing a button on the phone or on a Bluetooth headset. A benefit of this approach is that the user can leave the phone in their pocket leaving both hands free for using a cane¹⁴ and recognizing tactile¹⁵（触觉的） landmarks.

"This is a very cool mix of disciplines, using the user as a sensor¹ combined with sophisticated localization algorithms from the field of robotics," Folmer, of the University's Computer Science Engineering Human-Computer Interaction Lab, said.