RO-MAN 2026 study: Autistic adults want social robots as configurable rehearsal partners

A new paper from Yuval Zohar, Mordi Benhamou, and Guy Laban, accepted at the 35th IEEE RO-MAN 2026 conference, tackles a glaring gap in social robotics: most autism-focused robot research centers on children and rarely includes autistic adults as co-designers. The team conducted an online focus group and co-design session with five autistic adults, using a 90-minute structured activity on a shared digital whiteboard. Through reflexive thematic analysis of the resulting verbal and visual data, they distilled seven design requirements for robots intended to support social-cognition training in real-world adult contexts.

The findings are a clear departure from the companion or live-assistant paradigm. Participants envisioned the robot as a private, configurable rehearsal partner—one that acts as a scaffold rather than a substitute, favors authenticity over comfort, and gives the user personalized, user-controlled feedback. The design requirements also include accommodating gaps in emotional self-awareness, respecting privacy and contextual boundaries, supporting rehearsal for ambiguous interpersonal situations, and allowing the robot's identity, form, and expression to be customized. This implies building robots that fade into the background as the user gains independence, rather than becoming a permanent crutch. For designers and researchers, the study offers a concrete, adult-led blueprint for social-cognition training tools that actually match the everyday challenges autistic adults face, such as conversational timing and implied emotional meaning.