Room reconstruction and navigation using acoustically obtained room impulse responses and a mobile robot platform

This work explores the design and effectiveness of a robot that uses a combination of active sonar and a pseudo-random acoustic signal to navigate and reconstruct an unknown environment. The robot sends the pseudo-random signal into the environment and records the resulting response. This response is processed to gain information on the robot's immediate surroundings. Previous work done in this area focused on the recording and processing aspect to determine the location of a sound source or multiple sound sources. We apply similar algorithms to localize what are known as virtual sound sources. Virtual sound sources are created when sound from a sound source reflects off of a surface, such as a wall or object, and are recorded by a receiver. The recorded reflected sound is commonly known as an echo. A virtual sound source is placed at the location where the sound incident to the receiver would have originated from had no reflection taken place. By placing the real sound source and receivers on the same platform, if we can accurately localize the generated virtual sound sources, we can compute the path the sound wave took and localize the surfaces off of which the sound wave reflected. We remember these surface locations to generate a map for the robot to use when navigating through the unknown environment. Finally, we present a method to improve the accuracy of the baseline virtual sound source localization algorithm by using quadratic interpolation