Since the establishment of the paradigm centered on convolutional neural networks and recurrent architectures in 2020, the fundamental backbone networks for audio-visual navigation have not undergone essential changes for over five years, making them inadequate for efficiently representing dynamic multimodal sequences. This paper proposes Samba (A Hybrid Mamba for Audio-Visual Navigation). It employs the adaptive selection-enabled Mamba State Encoder (M-SE) to replace conventional GRUs for temporal aggregation, and constructs an Audio Mamba Encoder (AME) to address the limitations of convolutional operators in capturing global time-frequency dependencies in spectrograms. Experiments demonstrate that Samba exhibits exceptional generalization performance when facing unheard sound sources and unseen scenes. On the Matterport3D dataset, it improves the navigation success rate (SR) by 11.3% compared to existing state-of-the-art models, and the performance gain is even more pronounced on the Replica dataset, which features finer scene structures. Such modernized architectural reconstruction unlocks stronger embodied representation capabilities at a lower computational cost, thereby providing a highly robust technical pathway for paradigm evolution in the field of audio-visual navigation.
Blogger's Review: The introduction of Samba signifies a vital advancement in audio-visual navigation technology, particularly in handling dynamic and complex scenes. Its innovative encoder design not only enhances performance but also improves computational efficiency, showcasing the potential for future multimodal interactions. The strong generalization ability in unseen scenarios opens up new possibilities for practical applications.