What is it about?

Using existing tools from the field of autism, we developed an interactive robot that can engage with humans in structured interactions. The robot is programmed to respond to given stimuli in different ways emulating different levels of severity of autism. The level of severity along several dimensions is customizable by the user. Note that we are not attempting to simulate cognitive aspects of autism, but rather using a "shallow model" of behaviors that typically appear in the context of a standardized interaction widely used by clinicians. Our robot can be used to better train autism professionals at diagnosing children, as well as serve as an educational tool for both individuals with autism and the general public.

Featured Image

Why is it important?

Autism clinicians have been shown to display high levels of subjectivity when diagnosing for autism through behavioral observation. This can lead to serious consequences for patient's lives including false diagnosis and as a result inappropriate treatment. One of the reasons for this subjectivity is that current training methods do not incorporate interactive, procedural training but only relies on watching a large number of videos. Many of the clinicians in training are not prepared to engage in some of the highly structured tasks required during the diagnostic procedure. Robots could offer a way to complement diagnostic training by allowing clinicians to train the interactive procedure with a robot before administering it on real children.


This work was developed with the help of therapists who have been working in the field for many years. We were delighted to see how their initial skepticism of "simulating autism" turned into admiration as to how this tool was specifically designed to help them train the diagnostic procedure, allowing for repeatability and control, which is impossible with children.

Kim Baraka
Carnegie Mellon University

Read the Original

This page is a summary of: Interactive robots with model-based ‘autism-like’ behaviors, Paladyn Journal of Behavioral Robotics, March 2019, De Gruyter,
DOI: 10.1515/pjbr-2019-0011.
You can read the full text:




The following have contributed to this page