No data?

摘要

Since the autism spectrum disorder (ASD) diagnosis relies solely on behavioral observations by experienced clinicians, we investigate whether parts of this job can be autonomously performed by a humanoid robot using sensors available on-board. We have developed a robot-assisted ASD diagnostic protocol and propose the Partially observable Markov decision process (POMDP) framework for such protocol to enable the robot to make decisions about next actions. To obtain some of the parameters in our POMDP models we survey experienced ASD clinicians and encode their knowledge of children»s reactions in observation probabilities of POMDP models. We evaluate our approach in a small study with four children (two children with ASD, two typically developing children).