In 2001, students at the University of Pittsburgh and Carnegie Mellon University developed a Nursebot called Pearl that would roll around offering snacks to old folks in retirement homes, offer guidance when needed, and remind patients to take their medication. That robot project grew out of a couple earlier robots by CMU called Florence and FLO designed to study human-robot interaction. Now, almost a decade later, Carnegie Mellon University is at it again with SnackBot to deliver tasty treats to people in office buildings.
So what has changed over the past 10 years? To begin with, SnackBot relies on verbal and non-verbal communication, and uses the current standard in speech synthesis to speak in a more natural voice compared to Pearl. Pearl also required patients to interact through a touchscreen interface, something SnackBot doesn’t need thanks to its speech recognition. Pearl’s face had roving eyeballs, detached eyebrows, and bright red lips which, while more expressive, were somewhat off-putting compared to SnackBot’s simplistic design. For the time being, SnackBot should be fine for studying human-robot interactions.
SnackBot comes equipped with a laser range finder and cameras which will be used to research navigation, path-planning, and people/face detection in an office environment. The project aims to develop a robot that can recognize a person it has interacted with before, learn and recognize objects, and move discreetly in a crowded space, among other things. The robot could then be easily modified to perform as a security guard patrolling the office at night, or as a guide robot. In the following video, SnackBot answers the age-old question, “Do androids dream of electric sheep?“
- SnackBot project homepage
- CMU / Pitt Nursebot Project homepage
- Erik Glaser’s Industrial Design Portfolio
- SnackBot @ Flickr