A method for obstacle-range estimation based on echo-location using high-end sound with applications
摘要
Efficient mapping algorithms have been implemented in different fields of robotics applications. Some of those applications are in the mapping of environments with difficult accessibility, the planning of service or emergency assistant robots, or as supporting tools in helping people with disabilities. In those examples, the essential task of a mapping algorithm is to detect obstacles which may interfere in motion paths. One of the most widely used characteristics in pulse-echo range estimation is the time of flight. However, the ultrasonic frequencies (range from 30 to 120 KHz) are poorly suitable for the long distances faced in in-door applications. In view of the importance of developing efficient mapping algorithms, the present manuscript reports on an efficient method based on high-frequency waves in order to range static objects under a controlled environment. In the present approach, the signal acquisition is made out-of-band of the processing algorithm. This process is carried out by using an active speaker to throw the control ping sound. The resulting audio is captured through a unidirectional dynamic microphone Shure® SV100. The audio signal processing is based on selective filtering via Gabor models and wavelet decimation de-noising. The proposed method is compared against two common methodologies for range estimation in robotics based on ultrasonic and infrared sensor. The current methodology demonstrated a competitive and, in some cases, better performance than the standard methods.