Robotics could be a big step in autism therapy, but stumbling blocks remain

Using interactive robots to help children with autism has met with some success, but concerns remain about the ethics of the therapy and its high cost, writes Clara Chow

PUBLISHED : Tuesday, 25 February, 2014, 11:32am
UPDATED : Tuesday, 25 February, 2014, 11:32am

Seven years ago, robotics researcher Dr John-John Cabibihan was babysitting his 15-month-old daughter Marie, when, on a whim, he decided to see how she would react if he plonked a robot in front of her. Even though Marie didn't have any developmental issues, he watched as she interacted curiously with the robot. Cabibihan's thoughts turned to using robots as a tool for working with children who do have developmental issues, specifically autism spectrum disorders (ASD).

People with ASD often have difficulty communicating, fail to respond to social cues, tend to repeat particular actions (walking in a specific pattern for example) and sometimes become preoccupied with certain objects.

Some of our students interact with NAO better and more comfortably than their teachers do
Olivier Joubert 

Cabibihan thought "that some day it might be possible for data to be obtained and be processed for the diagnosis and therapy of children with autism. And that a robot companion could be a non-intrusive tool to do this."

While he was at the National University of Singapore, Cabibihan and other researchers surveyed the use of robots in treating autistic children. The results, which were published last year in a paper titled "Why Robots?" in the International Journal of Social Robotics, showed how therapists introduced robots to children with autism and supervised them in free play as they examined, held eye contact with and imitated the facial expressions on the robot - sometimes even responding with glee.

The paper points out that social robots, which tap into the fascination children have with the machines, could help in diagnosing autism by tracking gaze patterns or checking the response to repetitive actions. Developments in social robotics are "getting to be exciting every year", Cabibihan says.

Cabibihan, who now teaches at Qatar University, is one of a number of scientists working on using robots as a way of connecting with children with autism.

ASK (autism solution for kids) NAO is an initiative by French company Aldebaran Robotics to customise its humanoid robots for use in classrooms with children with autism. It has been used by more than 25 institutions worldwide since it was officially launched in last year. And in Belgium, a team at Vrije Universiteit Brussels, led by of professors Dirk Lefeber and Bram Vanderborght, has reported encouraging results with Probo, a huggable robot.


"Robots are able to provide a safe, simplified, predictable, interactive environment that can be repeated in the same format until the learning process is realised. The complexity of interaction can be controlled and gradually increased to respect different developmental levels. In this context, esearchers demonstrated that children with ASD are more responsive to feedback, including social feedback, when it is provided by technological means rather than by a human," Vanderborght says.

ASK NAO is being tested in four schools - three in the United States and one in Britain - and "many are waiting for delivery of their NAOs to join the adventure", says Dr Olivier Joubert, Aldebaran's autism business unit manager.

An article published in The Telegraph this month examined ASK NAO's partnership with Topcliffe Primary School in Birmingham, England, where autistic children played games, sang, danced and even read a self-penned suspense story to the robot.

"Some of our students interact with NAO better and more comfortably than their teachers do," says Joubert.

"They also give him a personality, a name, talk to him as they would a person; NAO becomes really a friendly mate for them, not only a tool in the classroom. Because of this, they become more and more confident to interact with him. It increases their self-esteem in a way that they are keen to try new things and learn even more and then take this learning outside the classroom into the world."

A primary objective of ASK NAO, Joubert says, is to enable parents to one day duplicate the set-up in their own homes - to "create a continuum between educational lessons at school and the ones able to be taught at home".

An ASK NAO package, which includes the robot, access to Aldebaran-developed applications and an online interface, and a one-year warranty on hardware, costs about US$10,000.

While there is much to be hopeful and excited about using social robotics in treating children with autism, researchers believe it will be a while before they become a mainstream option. Indeed, some experts are advocating caution and taking a long, hard look at the ethics involved in embracing robotics.

Vanderborght points to data which estimates that each person diagnosed with autism spectrum disorders costs €2.45 million (HK$25.8 million) in societal costs throughout his or her life, most of which is incurred during childhood. Thus, the eventual goal is to produce robots that exhibit "supervised autonomy" - to be able to operate for relatively short periods of time without human control, so as to reduce the workload of therapists - but with control reverting to human therapists whenever circumstances require.

"The more autonomous the robot, the less control therapist, parents, robot operator or designer has over the robot-child interaction. This then raises the issue of who is responsible for the robot's actions and behaviour," says Vanderborght.

He cites a Eurobarometer study from 2012, which showed that 60 per cent of European Union citizens said that robots should be banned from the care of children, the elderly and people with disabilities.

In an ongoing survey of parents and caregivers connected with autism, Vanderborght and his team found 47 per cent of respondents agreeing and 39 per cent strongly agreeing that it is ethically acceptable for social robots to be used in therapy for children with autism.

That approval level dropped, however - with only 19 per cent agreeing and 8 per cent strongly agreeing - when it comes to replacing human therapists with social robots to teach skills to children with autism, leading the team to conclude that supervised autonomous robots are the way to go.

Joshua Diehl, a psychologist at the University of Notre Dame's William J. Shaw Centre for Children and Families who studies how robots can be used in autism-related therapy, expresses concern that robots are being marketed for children with ASD without clinical research support.

"I still have this concern. I wanted to objectively study the issue," he says.

He presented his findings at last year's International Meeting for Autism Research in San Sebastian, Spain. While he and his team found that children showed greater gains in learning social skills when a robot co-therapist was present, he also found that results varied widely from child to child.

"It's not enough that children like the robots, they have to show greater improvement than in existing therapies, and it has to be cost effective. I do not think that these points have been consistently demonstrated in research to date," says Diehl.

Given the high costs involved, Diehl suggests that families should not pursue social robotics as a primary course of intervention unless it is part of a research study or an already established school programme.