Research On Sound Source Localization Technology Of Humanoid Intelligent Robot Based On Microphone Array

Sound source localization is a basic problem in many areas. The research onsound source localization technology has always been a research focus, it related tomany technology fields, such as acoustic, signal detection, digital signal processing,and so on. It also has very broad application prospects. At present, the application ofthe technology in humanoid intelligent robot is not a lot yet. However it as animportant way of human-computer interaction, it can be able to combine with theinteraction of vision perfectly and to make up for its shortcomings.By researching and comparing a variety of sound source localizationtechnology, it can get that the source localization technology based on TDOA(Timedelay of arrival) values is more applicable. The principle of this technique is notvery complicated, it without a large amount of calculation, and compliment thistechnique easily. Currently, the research on this technology is relatively mature, ithas a large number of applications. Thus it is a good choice for the research of thisproject. By using TDOA technology, the work needs to be done in two stages: Inprevious phase, it uses delay estimation algorithms estimating the TDOA values thatsound source signal arrives at each microphone, that is the time difference; in thesecond stage, according to the TDOA values obtained, it uses spatial positioningmethod to locate the sound source with geometry relationship of the array and thespatial relationship. Thus whether can get high-precision values of TDOA is veryimportant for the entire process completed correctly. Traditional GCC algorithm is akind of classical approach of sound source localization technology based on TDOA.It can locate the sound source quickly, but its precision and accuracy of positioningis not very high and it vulnerable to the impact of environmental and noise. In thispaper, it makes some improvements to the shortcomings of traditional GCC delayestimation algorithm and improves the accuracy of the delay estimation. Theimprovement on GCC delay estimation algorithm is the key and core of this article.In this study, the microphone array uses a two-dimensional planar structure.Combined with the time delay estimation algorithm and spatial positioning methodcorrespondingly, a sound source localization system was designed and implementedat last. The microphone array was consisted by four microphone modules. Then the construction of this system was complete together with the data acquisition card andPC. The sound source localization algorithms were implemented on the PC andcompleted the software program of system, the system was installed on the robot atlast. After completed the system, a mount of experiments were conducted to test andvalidate the system. The results of experiments validated that it was feasible toimplemente sound localization system on the humanoid intelligent robot, the systemcan achieve a certain localization effect and control the robot’s motions with thepositioning results. Therefore it achieves the human-robot interaction friendly.