Preliminary investigation of obstacle recognition via smartphone active sound sensing for pedestrian safety

To ensure the safety of the pedestrians that use a smartphone while walking, numerous smartphone applications have been developed. The most common method is to use the back camera of the smartphone to record the video in real-time to recognize obstacles such as approaching vehicles and people. However, this imposes problems such as limited detection ranges, and poor performance in the dark. To overcome these problems, we design a method that can recognize common obstacles such as vehicles, trees, signposts, etc. using smartphone active sound sensing. We mimic the echolocation of bats and emit sine waves and sweep signals from the smartphone and record the reflected waves. We exploit the spectral and spatial characteristics of the stationary and non-stationary obstacles and recognize them, making our method adaptable to numerous applications such as assisting visually impaired people, managing smartphone alert levels, and providing information regarding how crowded the sidewalk is.