Social robots for autism
A robot is predictable and endlessly patient, which can make it a useful partner in therapy for children with autism. A robot in a playroom also gets dropped, shaken and thrown. The studies here ask how to make a small robot safe to throw, and how it can tell from its own sensors that a meltdown is coming.
What if the robot is thrown?
A 2018 study warned that a small robot thrown at a head during a meltdown could cause tissue injuries, and in extreme cases a concussion. So the lab built a low-cost rig that throws objects at a dummy head fitted with an accelerometer, and measured the head injury criterion, the 3 ms criterion and peak acceleration. Pick a toy and a speed.
The study's safest toy was a 0.3 kg cube or wedge.
The soft material tests wrapped a 0.4 kg cylinder in Ecoflex, Dragon Skin or clay, 1 to 5 mm thick.
Sources: the rig, HardwareX 2019; shape and mass, 27 Taguchi runs; soft materials, all in 2019. Energy is computed from the mass and speed you pick. The head's motion is drawn for illustration.
A toy that knows it is being thrown
An accelerometer inside a toy can tell a gentle pick-up from a throw. The lab recorded children playing with a stuffed panda, a stuffed robot and a toy excavator, and trained models on six behaviours: idle, pick up, shake, hit, drop and throw. Later work moved the sensors onto the child, as a wearable that also reads heart rate.
In free fall the reading drops to zero, then the landing shows up as a sharp spike.
Sources: RO-MAN 2019, RO-MAN 2021, IJSR 2022, Robotics 2023, and the toy dataset. Each study used its own recordings, so the bars are not one benchmark. The acceleration traces are drawn for illustration.
Fast enough to answer
Five machine learning methods were compared for a robot that reacts as it happens. XGBoost was the best at 90% accuracy, with an answer in 0.004 s. The accelerometer told the model more than the gyroscope did.
The heart rate gives it away
On a wearable, body signals beat motion signals, and heart rate was the strongest single feature. RMSSD, a measure of how much the time between heartbeats varies, tracked the challenging behaviours. This work became a US patent.
How fast should the robot answer?
When a child shakes the toy, the robot answers with a sound. If the answer comes too late, the child may not connect it to what they did. Choose a delay and shake the toy.
Quick enough to feel like an answer.
A robot answered a child's action with a sound after 0.5, 1.0 or 1.5 s. For shaking, the delay made a significant difference, and the study concluded that replies slower than 1.0 s may hurt comprehension.
Surveys, teachers and trains
Why robots?
A survey of the roles robots play in autism therapy and what they bring to it. With 523 citations, it is the most cited paper in this chapter.
Asking the teachers
Six special education teachers were interviewed about robots in their classrooms. They were skeptical, saw robots doing only repetitive tasks, and expected them to be bullied.
Thomas and friends
Many children with autism love Thomas the Tank Engine, and parents report gains in social skills and speech. The paper asks why, and what that means for robots that tell social stories.
Sensing for screening
A review of six kinds of sensors used to screen for autism and support therapy: eye trackers, movement trackers, skin conductance, touch, voice and sleep.
Robotic trains
The train went from idea to a therapy and teaching tool. Its design later became a US patent, with an arm that loads coal and a locomotive that blows bubbles.
Papers and grants
23 papers
- 2023Heart Rate as a Predictor of Challenging Behaviours among Children with Autism from Wearable Sensors in Social Robot Interactions
QURobotics, 25 citations - 2023Editorial: Technologies to support the diagnosis and therapy of individuals with autism
QUFrontiers in Psychiatry, 1 citation - 2022Real-Time Social Robot’s Responses to Undesired Interactions Between Children and their Surroundings
QUInternational Journal of Social Robotics, 11 citations - 2022A Framework for Assistive Social Robots for Detecting Aggression in Children
QULecture Notes in Computer Science, 1 citation - 2021Detection of Challenging Behaviours of Children with Autism Using Wearable Sensors during Interactions with Social Robots
QU2021 30th IEEE International Conference on Robot & Human Interactive Communication (RO-MAN), 23 citations - 2021Datasets for recognition of aggressive interactions of children toward robotic toys
QUData in Brief, 4 citations - 2020Influence of Reaction Time in the Emotional Response of a Companion Robot to a Child’s Aggressive Interaction
QUInternational Journal of Social Robotics, 19 citations - 2019Recognition of Aggressive Interactions of Children Toward Robotic Toys
QU2019 28th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN), 16 citations - 2019Safety experiments for small robots investigating the potential of soft materials in mitigating the harm to the head due to impacts
QUSN Applied Sciences, 13 citations - 2019A low-cost test rig for impact experiments on a dummy head
QUHardwareX, 9 citations - 2019Influence of the shape and mass of a small robot when thrown to a dummy human head
QUSN Applied Sciences, 8 citations - 2019Data on the impact of objects with different shapes, masses, and impact velocities on a dummy head
QUData in Brief, 4 citations - 2019Data on the impact of an object with different thicknesses of different soft materials at different impact velocities on a dummy head
QUData in Brief, 3 citations - 2018Head Impact Severity Measures for Small Social Robots Thrown During Meltdown in Autism
QUInternational Journal of Social Robotics, 17 citations - 2018Robotic Trains as an Educational and Therapeutic Tool for Autism Spectrum Disorder Intervention
QUAdvances in Intelligent Systems and Computing, 17 citations - 2018Social Robots and Wearable Sensors for Mitigating Meltdowns in Autism - A Pilot Test
QULecture Notes in Computer Science, 16 citations - 2016Sensing Technologies for Autism Spectrum Disorder Screening and Intervention
QUSensors, 68 citations - 2016Culture as a Driver for the Design of Social Robots for Autism Spectrum Disorder Interventions in the Middle East
QULecture Notes in Computer Science, 4 citations - 2015Social Robots
QUProceedings of the 3rd 2015 Workshop on ICTs for improving Patients Rehabilitation Research Techniques, 17 citations - 2015Thomas and friends: Implications for the design of social robots and their role as social story telling agents for children with autism
QU2015 IEEE International Conference on Robotics and Biomimetics (ROBIO), 6 citations - 2015Autism in the Gulf States: Why social robotics is the way forward
QU2015 5th International Conference on Information & Communication Technology and Accessibility (ICTA), 5 citations - 2013Why Robots? A Survey on the Roles and Benefits of Social Robots in the Therapy of Children with Autism
NUSInternational Journal of Social Robotics, 523 citations - 2012Cute and soft
NUSProceedings of the Workshop at SIGGRAPH Asia, 20 citations
Grants
- 2020Smart, Snap-on Device for Detecting Aggressive Behaviors in Children with Autism during Meltdown Events
US$150,137 · Qatar University · Lead Principal Investigator - 2019Safe and Adaptive Social Robots for Children with Autism (Phase II)
US$5,492 · Qatar University · PhD Supervisor - 2018Safe and Adaptive Social Robots for Children with Autism
US$5,492 · Qatar University · PhD Supervisor - 2014Social Robotic Toys for Children
US$12,357 · Qatar University · Sole Investigator - 2012Robotics for Autism Assessment
US$141,376 · National University of Singapore · Lead Principal Investigator