Research On Sound Source Location Technology Based On Mobile Robot Platform

With the rapid development of science and technology,intelligent products are widely used in many fields such as daily life,scientific research,measurement and forecasting,and industrial production.Signal acquisition and processing is an indispensable part of intelligent products as a means of intelligently collecting external information.The sound source signal has many advantages such as strong propagation ability,strong anti-interference ability,and strong diffraction capability,etc.Therefore,the sound source localization technology has gradually become a research hot spot.Unfortunately,the process of acquiring the sound source signal is subject to unavoidable external interference factors,such as environmental noise and reverberation,which may cause serious problems such as insufficient positioning of the sound source positioning system and poor real-time performance,etc.In this thesis,some existing microphone array models and sound source localization algorithms are studied in depth.A nonlinear filtering technique is used to estimate the sound source location with the aid of the mobile robot platform.The specific work contents are given in the following chapters:(1)The research status of sound source localization algorithm is summarized in Chapter I.The method of sound source localization based on time delay estimation algorithm is studied due to the consideration of experimental cost and product cost performance.(2)Before the sound source localization procedure,the sound signal collection method and processing method were recalled in detail in Chapter II to lay the foundation for a better quality sound signal.Two research models of sound sourcesthe near-field sound source model and the far-field sound source model are compared with each other.(3)The topology of the microphone array is studied in depth in Chapter III.According to the existing microphone array topology,a six-element planar hexagonal microphone array structure is designed,and the space of the six-element planar hexagonal microphone array is derived with geometric positioning method.(4)In Chapter IV,more common generalized cross-correlation time delay estimation methods and miinimum two-dimensional adaptive filtering time delay estimation algorithms are studied in depth.The comparison shows that,the delay estimation with least squares adaptive filtering has better anti-reverberation capability,and thus is chosen as the initial value of the state variable of nonlinear filtering sound source localization technology.Moreover,a method based on nonlinear filtering technology for sound source localization,which simultaneously estimates the delay between the microphone array elements and the position of the sound source target is proposed in this chapter.(5)To aid the experiment verification of the positioning accuracy and real-time performance with the proposed methods for sound source localization,a mobile sound source localization robot system was constructed.It can adjust the posture of the microphone array by adjusting the posture of the robot while the sound source is being tracked,so that the sound source target is always within the optimal collection angle range of the two-dimensional microphone array,leading to the improvement of the the sound source localization precision.