マルチロータヘリコプタによる識別情報を考慮した複数音源位置推定

抄録

In this paper, it is considered that a multirotor helicopter with a microphone array localizes the sound sources on the ground. Since the noise generated by rotors distorts the target signal, the estimated sound source direction is uncertain. Besides, the helicopter perceives signals from multiple sources because of its wide range of view from the sky, which makes it difficult to fuse multiple observations. This paper proposes multiple target detection by using data association technique with classification measurements under the existing of false detection. The classification measurements with SVM is integrated into the data association to suppress false detections and false associations. The effectiveness of the proposed approach is validated through flight simulation.