The Autism Society Event and Education Recordings Archive

Autism, or Autism Spectrum Disorder (ASD), is a developmental disorder that can cause social, communication, sensory, and behavior challenges. As for many other developmental disorders, early diagnosis and intensive intervention are critical to lowering treatment cost and improving prognosis. Unfortunately, it often requires enormous professional experience and observation to recognize symptoms of ASD. Intensive interventions are also known to be expensive. Hence there has been a great interest in technology that can recognize and measure symptoms of ASD and even facilitate interventions.

Restricted and Repetitive Behaviors (RRB) are one of the core symptoms and diagnostic criteria of ASD. Although no single type of RRB is unique to autism, a pattern of multiple types of RRB can be an early indicator of ASD. RRBs are also a frequent target of behavioral interventions. Among them, stereotypical motor movements (SMM) refer to repetitive, non-functional motor behaviors (e.g., body rocking and hand flapping). Sensors, such as wearable accelerometers and video cameras, can help recognize and measure SMMs, which provide valuable data for identifying markers of ASD and assessing treatment progress.

However, not all RRBs exhibit recognizable acceleration or visual patterns. For example, when playing with toys (e.g., building blocks), many children with ASD tend to play in restricted and repetitive ways (e.g., stacking). Smart objects with embedded wireless sensors can fill this gap by recognizing movement patterns of the objects being manipulated by an individual.

The remaining problem is when a RRB is identified, it still requires human professionals to implement interventions. To our knowledge, few technology can facilitate or even implement behavioral interventions automatically. Our work pushes forward such a technology to not only recognizing and assessing RRBs, but also delivering behavioral interventions right during play. Instead of SMMs, we can address more subtle forms of RRBs such as sameness in object manipulation.

Our technology, named Guided Play, is an intelligent play companion for young children that functions as a behavior co-therapist. It observes a child's play, measures behavior variability, detects repetitive patterns, and joins the play to increase behavior variability and repertoire by implementing contingencies and incorporating prompts. This presentation introduces the guided play technology and our initial evaluation of it.