Scrabble-playing robot aimed at refining human interaction
A multidisciplinary team has developed Victor the Gamebot, a robot that regularly plays games of Scrabble with students at Carnegie Mellon University.
Developed under the direction of Reid Simmons, research professor in CMU’s Robotics Institute, Victor the Gamebot is the latest in a series of so-called social robots that anticipate a time when people and robots will interact routinely.
Simmons and fellow researchers are using Victor to better understand the variables required for people to engage with a robot and enjoy the process too.
‘We believe that for autonomous robots to be accepted, they will have to conform to the social conventions of people, rather than the other way around,’ Simmons said in a statement.
The researchers are investigating whether changes in mood or emotions affect the desire to interact with robots and how personalisation, such as the robot remembering a person’s play from previous games, might affect the willingness to interact over time.
Victor’s torso is topped with a mobile head on which a video screen displays its animated face, designed by Anne Mundell, associate professor of scene design.
The machine is seated at one end of a table-size touchscreen on which the Scrabble board and virtual tiles are displayed; three seats are available for human players.
Victor has no arms, but is able to move tiles electronically whilst opponents move their virtual tiles with their fingers. Victor has a voice to address his opponents and people can converse with Victor using keyboards.
The gamebot operates for a couple of hours each day in a public space at the University, and a research team member is present to address any problems. As reliability improves, the goal is for Victor to be available for games anytime, without need of supervision.
The CMU team note that Victor is not a strategic player, paying little heed to double- and triple-word scores, and his 8,600-word vocabulary is modest compared with the Official Scrabble Player Dictionary’s 178,000 words.
The research was supported in part by a grant from the Qatari National Research Fund.