Searcharxiv⌕ Search
Home/Archive/arXiv

arXiv · 1804.03372

Realtime Active Sound Source Localization for Unmanned Ground Robots Using a Self-Rotational Bi-Microphone Array

Deepak Gala·Nathan Lindsay·Liang Sun
View original source ↗Read source document ↗Export metadata ↓

Abstract

This work presents a novel technique that performs both orientation and distance localization of a sound source in a three-dimensional (3D) space using only the interaural time difference (ITD) cue, generated by a newly-developed self-rotational bi-microphone robotic platform. The system dynamics is established in the spherical coordinate frame using a state-space model. The observability analysis of the state-space model shows that the system is unobservable when the sound source is placed with elevation angles of $90$ and $0$ degree. The proposed method utilizes the difference between the azimuth estimates resulting from respectively the 3D and the two-dimensional models to check the zero-degree-elevation condition and further estimates the elevation angle using a polynomial curve fitting approach. Also, the proposed method is capable of detecting a $90$-degree elevation by extracting the zero-ITD signal 'buried' in noise. Additionally, a distance localization is performed by first rotating the microphone array to face toward the sound source and then shifting the microphone perpendicular to the source-robot vector by a predefined distance of a fixed number of steps. The integrated rotational and translational motions of the microphone array provide a complete orientation and distance localization using only the ITD cue. A novel robotic platform using a self-rotational bi-microphone array was also developed for unmanned ground robots performing sound source localization. The proposed technique was first tested in simulation and was then verified on the newly-developed robotic platform. Experimental data collected by the microphones installed on a KEMAR dummy head were also used to test the proposed technique. All results show the effectiveness of the proposed technique.

Explore related subjects

cs.SDcs.ROeess.ASeess.SP

Keep this discovery

BibTeXRIS

Deepak Gala, Nathan Lindsay, Liang Sun. 2018-04-10. Realtime Active Sound Source Localization for Unmanned Ground Robots Using a Self-Rotational Bi-Microphone Array. https://arxiv.org/abs/1804.03372

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Xiaomi-CocktailASR-1 Technical Report

Recently, large language model (LLM) based ASR models have achieved significant progress, yet they generally lack support for multi-speaker scenarios, where the cocktail party problem remains a critical bottleneck for further advancing ASR. Existing TS-ASR methods, including end-to-end architectures with speaker embeddings and latest LLM-based explorations suffer from degraded single-speaker performance and the inability to reject when the target speaker is absent. In this paper, we propose Xiaomi-CocktailASR-1, an LLM-based end-to-end TS-ASR architecture. By utilizing reference speech as voiceprint prompts, it directly transcribes the target speaker's speech without requiring speech separation. Xiaomi-CocktailASR-1 maintains competitive performance in single-speaker scenarios, comparable to mainstream ASR models. It also features a negative sample rejection capability, outputting empty text when the target speaker is absent from the mixed speech. Additionally, Xiaomi-CocktailASR-1 supports a Chain-of-Thought (CoT) reasoning mode to provide explicit reasoning steps. Extensive experiments on various synthetic and real-world multispeaker benchmarks demonstrate that Xiaomi-CocktailASR-1 achieves state-of-the-art performance, effectively addressing the cocktail party problem through a unified architecture that balances multispeaker and single-speaker recognition accuracy, along with rejection capability.

cs.SD↗
arXiv2026

EConv-TasNet: Efficient Conv-TasNet for Effective Speech Separation

Conv-TasNet has served as a strong baseline for time-domain speech separation, and many studies have extended it with advanced architectures such as dual-path networks, U-Nets, and attention mechanisms. However, these methods often introduce high computational cost and complexity, limiting their deployment in resource-constrained scenarios. To address this issue, we propose eConv-TasNet, an efficient variant of Conv-TasNet that improves both effectiveness and efficiency without relying on resource-intensive modules. The proposed model consists of a group-wise early-splitting (GES) module and a multi-group feature aggregation (MGFA) module. GES generates discriminative speaker embeddings at intermediate stages, while MGFA progressively aggregates these group-level representations for refined mask estimation. Experimental results show that eConv-TasNet reduces model size by 22.4%, accelerates inference by 18.9%, and improves SI-SNRi by 14.0%-28.0% across three public benchmarks. Moreover, it achieves competitive performance compared with state-of-the-art methods while requiring significantly fewer parameters and lower inference cost. These results demonstrate a favorable efficiency-effectiveness trade-off for edge deployment.

cs.SD↗