When children don’t feel like talking with others, using humanoid robots as go-betweens might help, researchers say.
Increasingly, scientists are devising what are called “social robots,” machines such as robotic pets that are designed to interact with people. One such robot invented by researchers in England is the humanoid KASPAR, which stands for Kinesics and Synchronization in Personal Assistant Robotics.
First built in 2005, KASPAR has a child-friendly size, posture and clothing to help it work with kids, including autistic children. Motors in KASPAR’s face, neck and arms create facial expressions and gestures and move its limbs and body like a human. Tactile sensors in the hands, feet, chest, arms and face enable the robot to respond to touch — for instance, giggling when tickled and looking sad when hit.
KASPAR’s simplified face looks strange to some, they explain, but its design is aimed at making autistic children feel comfortable, as they prefer a minimum of facial expressions and features.
From autism to traumatic settings
Scientists reasoned that robots such as KASPAR might help people interview children better than they could do it alone when it came to emotionally sensitive situations such as interactions with police, healthcare and social services.
"In a healthcare setting, this could be finding out about how a child acquired a particular injury. In a police or social services setting, this could be finding out about specific events such as cases of abuse and, more importantly, if the child is in any immediate danger," says roboticist Luke Wood at the University of Hertfordshire in England.
"Children can sometimes feel intimidated when talking to a person in an authoritative position and, as a result, will sometimes be unwilling to communicate," Wood says. "One advantage of using a small humanoid robot is that the child is less likely to feel intimidated, as the robot is not an adult authoritative figure. Therefore, we hypothesize that the child may be more willing to talk to a robot about sensitive topics."
Professional interviewers could in theory precisely control every aspect of the robot’s behavior, including facial expressions and body language, which could help children feel more comfortable. Such control is often very difficult even for professionally trained interviewers, especially if stressful or traumatic ordeals that children have undergone lead them to reveal surprising or shocking details that make it hard for interviewers to maintain their composure.
"We are trying to develop a tool to assist professional interviewers where conventional techniques are not working," Wood says. "We do not foresee a day where a fully automated robot is interviewing children, as this is such a skilled and delicate task and requires, for example, a deep understanding of human emotions and human nature in general. We are merely developing the robot as an interface, a robotic mediator, with the hope that it will help professional interviewers to communicate and reach children that are unwilling to communicate via conventional means."
To see how effective a robot interviewer might be, Wood and his colleagues compared how 22 children aged 7 to 9 responded to KASPAR versus an adult, Wood, when it came to an interview about an event known as Red Nose Day that had recently taken place at school to help raise money for charity. The children were unaware that a person was controlling KASPAR remotely.
Opening up to robot
The interviews began with a short introduction to establish rapport involving learning each other’s name, the child’s age and whether he or she had any siblings. The children were then asked for facts about the winner of the event and who the judges were, similar to how police would conduct an interview.
Analysis of their behavior and responses to questions revealed the replies the children provided KASPAR and Wood were very similar. Interviews with KASPAR lasted about a minute-and-a-half longer on average, mostly because technical limitations made KASPAR’s responses slow. However, this apparently did not lead to a disconnect with the children, who looked at KASPAR’s face about 40 seconds longer on average than they did Wood’s.
"Regardless of whom or what the children were talking to, the information they provided was very similar," Wood says. "This result would suggest that the children in our study may have viewed the robot as an equivalent interview partner, and this is an important step in establishing if robots could be useful tools in an interview scenario."
"When carrying out this exploratory study, it was perfectly possible that the children would either not respond at all to the robot due to the unfamiliarity of the situation, or that they would chat for long periods of time and make things up as they might in a fantasy-type play scenario," said he says. "The fact that the children responded to the robot in a similar way to which they did a human interviewer is very encouraging."
One limitation of this work is that KASPAR did not deal with children in the stressful scenarios that researchers might one day want the robot to work in.
"The children in our study were answering questions about a fun event that had taken place in the school," Wood says. "Further research would need to investigate how children would respond to questions of a more sensitive nature. Having said this, we believe that this is where the main advantages of a robotic interviewer may lie."
Wood’s team is now developing the robot further to make it more flexible and user-friendly for professional interviewers to try in more traumatic situations.
They detailed their findings online March 22 in the journal PLOS ONE. For videos of KASPAR in action, click here.
Top Image: KASPAR interviewing a child. Courtesy L.J. Wood et al.