Improving LEO Robot Conversational Ability via Deep Learning Algorithms for Children with Autism

摘要

The core symptom of children with autism is social difficulties. According to the research, one of the main psychological factors supposed to underlie in these difficulties is the lack or low levels of joint attention with the interaction partners. The use of robots in autism spectrum disorder (ASD) interventions has received a lot of attention in the last years. Robots can achieve high levels of effectiveness in interacting with children with autism. This paper presents robots that play several important roles and benefits in the interaction of children with autism. In the absence of dialogue corpus, we collected and integrated conversation data for children with autism. We present to use a neural network to build a robot dialogue system that generates answers freely without restrictions, and design robot movements to attract attention from children with autism. Most importantly, the robot will interact smoothly with autistic children without human intervention.