Herme Corpus
[row style=”margin: 20px 0px;”]
[column lg=”6″ md=”6″ sm=”12″ xs=”12″ ]
HERME is a Lego NXT robot that was augmented using open CV, Python and MAX/MSP to create a basic conversational robot. HERME sat in the Science Gallery in Trinity College for three months and talked to members of the public.
HERME @ THE SCIENCE GALLERY
Developed as part of the FastNet project, an SFI funded initiative, this experiment will allow the visitor to interact with a robot and gain an insight into how a robot might interact with them. The focus of the experiment will be on the non-verbal aspects of conversation to show how humans process a myriad of signals from their gestures and tone of voice— this is something robots are not as good at processing. Using a LEGO NXT (Mindstorms) robot as a platform for cameras and microphones the work will offer the visitor a robot’s-eye-view of a conversation with a human. The conversation data has been captured and used to create a collection of human-robot-interaction, as well as a substantial database of voices and speech from Science Gallery visitors.
Set Up and Method of Collection
The Herme Corpus, with more than 500 consent form signed, constitute a very promising material for the study of spontaneous human-robot interactions.
[/column]
[column lg=”6″ md=”6″ sm=”12″ xs=”12″ ]
Participants
Participants to the experiment where the visitors of the Science Gallery.
Timeline and Precis of corpus content
Under construction
[/column]
[/row]
Recordings
Timing and synchronisation
A database stores timepoints for all recordings, and offsets are used to synchronise the various media. The reference time used was the actual time of day, and all media timepoints are mapped to this ‘global’ time. For viewing several media simultaneously the offsets for each individual recoding’s ‘local’ start and stop times are used to synchronise playback.
| Title | Start Time | End Time | mp3 | 16kwav | 48kwav | Comment |
|---|
| Title | Start Time | End Time | mp3 | 16kwav | 48kwav | Comment |
|---|