Research on acoustic positioning system in robot audition

摘要

Robot auditory system was studied due to the advantages of non-line of sight and small data quantity, so the system was studied. Denoise background signal by EMD method, and then choose suitable IMF used for TDOA measured by GCC method. Substitute TDOA obtained into SI algorithm, then the position information of sound source could be get in near field. The influence factors on position error are TDOA errors and distance between robot and sound source. The maximum relative error is 6% with single channel contained TDOA error, and the maximum error is below 0.2m when all channels have TDOA error. The critical distance is 4m, and if it is less than 4m, the maximum error is below 0.2m, but it increases obviously when the distance more than 4m. EMD method can denoise background signal effectively, and the peak is obvious obtained by GCC method through EMD, the explicit boundary point on mobile robot by using different distance field was given, it can help the robot judge the sound source is in near field and far field. Robot auditory position system is feasible in simulation.