SpeeLo: Speech Interaction During Locomotion in Resource-Constrained Bipedal Robot

Abstract

In bipedal robots, achieving stable operation while simultaneously executing real-time locomotion control and computationally intensive speech processing is a critical design challenge. This paper proposes SpeeLo (Speech Interaction During Locomotion in Resource-Constrained Bipedal Robot), a distributed architecture and corresponding evaluation methodology for integrating speech interaction capabilities into bipedal robots with limited computational resources. In the proposed approach, a lightweight server deployed on the robot handles locomotion control and speech output, while speech processing is offloaded to an external client. Furthermore, dialogue history management enables context-aware response generation. Experiments with a physical robot quantitatively demonstrate that the proposed system has minimal impact on the locomotion control cycle and that incorporating conversational context improves response quality. These results confirm that SpeeLo enables safe and effective integration of speech interaction under constrained computational resources.
Additional material is available at https://mertcookimg.github.io/speelo