Research On Sound Source Localization System For Intelligent Robot Applications

Applying the human auditory system mechanism to the field of robotics not only complements the vision and enables the robot to flexibly handle audio-visual information in its environment,but also has a great value in the research of intelligent robots.As one of the main functions of the robotic auditory system,sound source localization has received extensive attention and research in recent years.It is the basis of other functions of the robotic auditory system,such as sound source separation,speech recognition and so on.Meanwhile,sound source localization plays a decisive role in the field of human-machine interaction(HMI,Human-Machine-Interaction)and artificial intelligence.Based on the previous researches on the sound source location system,this paper discusses the microphone array technology in depth and proposes a four-element cross array sound source localization system based on the Linux system Raspberry Pi ReSpeaker 4-Mic expansion board as the core array.By comparing the performance of the three types of algorithms such as beam-forming,super-resolution spectrum estimation,and time delay difference of arrival,Adopt the current mainstream time delay estimation(TDE,Time Delay Estimation)sound source localization algorithm as the final algorithm of the system,and tested the system performance and algorithm accuracy in different indoor and outdoor environments;finally set up a joint control step motor rotation system by DSP board ADI-21489 and STC89C52.Transplants the sound source localization system to the robot head,which realizes the function of sound source localization in the robotic hearing system and the robot head rotation towards the sound source.The system verification has achieved good results.The general work and content of this article can be divided into the following:1.Introduced the key technology of the microphone array in speech signal processing,and proposed a scheme based on the microphone array technology to realize the robotic auditory system;2.A quaternary cross-type microphone array based on the R-Speaker 4-Mic expansion board of Raspberry Pi is designed.The performance of the three types ofsound source algorithms such as beam-forming,super-resolution spectrum estimation,and time delay difference of arrival was compared by simulation.Based on the final algorithm of arrival delay difference to realize the system.Comparing the localization results of sound source localization algorithms such as GCC and LMS.An improved algorithm for time delay estimation based on least mean square error is proposed.Experiments show that the algorithm has higher precision time delay estimation in indoor reverberation scenes.3.The robot sound source localization system is designed.The four-element cross microphone array is mounted on the head of the robot,and the robot head is controlled by a stepping motor to realize the robot sound source localization and head rotation facing the sound source.The robot auditory system has achieved the target of "locating by sound".4.Summarize the main research content of this article,explore the functions of the robot auditory system implemented in this article,and put forward the shortcomings that need to be improved.