For human-robot and human-system speech-based communication, various researches have been actively studied, including automatic speech recognition (ASR), natural language processing, dialog management, and speech synthesis, by focussing on communication and/or interaction.

For example, ASR is a key component for natural and smooth human-robot interaction, but there are many obstacles to overcome in the real world, such as noise, interruption, barge-in, latency, and so on. Many techniques based on emerging deep learning technology have been proposed in the speech processing community to address these well-known problems, they have not been technically discussed in the context of human-robot communication, and thus these techniques have limitations to apply them to real human-robot communication scenarios due to lack of robustness and technological unmatch with a target application. This situation tells us the necessity of technical discussion considering application-specific requirements. 

The goal of this workshop is to identify real-world challenges and explore their solutions for more practical human-robot and human-system speech-based communication by sharing real-world application-specific problems, which are actually faced by speakers and audience in this workshop.

Submission Guidelines

We invite every author to submit one of regular paper (6-8 pages), short paper (2-4 pages), and extended abstracts (1-2 pages). A regular paper is considered as full paper, and a short paper is considered as a position paper or an ongoing topic. Extended papers are generally to provide an overview by researchers in industry who could not disclose details. All papers should be submitted using the RO-MAN paper format.

List of Topics

The possible topics include, but are not limited to:

• Robot audition
• Noise reduction and speech enhancement techniques
• Automatic speech recognition techniques
• Spoken language understanding techniques
• Dialog management system
• Speech, synthesis, text generation techniques
• Multi-modal processing techniques
• Non-verbal communication techniques
• Analysis, extraction and prediction of errors in human-robot communication
• System optimization for human-robot communication
• Software design for speech related human-robot communication
• EDI research topics related to speech such as sign language recognition and generation

Committees

• Yui Sudo (Honda Research Institute Japan Co., Ltd.)
• Kazuhiro Nakadai (Tokyo Institute of Technology)
• Katsutoshi Itoyama (Tokyo Institute of Technology / Honda Research Institute Japan)
• Muhammad Shakeel (Honda Research Institute Japan)

Venue

The conference will be held in Busan, Korea

Contact

Yui Sudo (Honda Research Institute Japan Co., Ltd.)

email:  yui.sudo@jp.honda-ri.com