}: According to the 2010 US census (www.census.gov/prod/2011 pubs/acsbr1 0-12. pdf) 2.8 million school-aged children (aged 5 to 17) suffer from some form of mental and/or physical disability. The Individuals with Disabilities Education Act (IDEA) guarantees such children equal access to education. However, children aged 1 to 5, while even more in need of early rehabilitation therapy and social interaction, do not receive the same attention. This proposal describes a plan for developing a new paradigm for pediatric rehabilitation. The paradigm involves robots that interact socially with special needs children, within a closed feedback control loop. Its components include new methods for modeling and representation of children motion, visual activity recognition, machine learning and behavior adaptation, and general-sum game-theoretic analysis of human-robot interaction. The research groups in the two collaborating institutions, University of Delaware (UD) and Johns Hopkins University (JHU), have produced preliminary results and evidence which suggest that the envisioned technical innovations in the domains mentioned above are feasible, and that physical implementation is possible. There are three intertwined research thrusts in the proposed activity, that address specific long-standing challenges in each area: 1. pediatric rehabilitation: develop a robot-assisted, pediatric rehabilitation environment which combines motor training with social interaction; 2. computer vision and perception: develop interpretable perception and machine learning algorithms for adaptive activity recognition; 3. robot control: develop the control and coordination algorithms to enable directed and purposeful human-robot interaction.
Each year, 10 000 infants in the U.S. develop cerebral palsy and 4 000 infants have birth defects of the spine and the brain. The motor delays for these infants have lifelong consequences. This gap in mobility is a significant secondary impairment. We propose a new portable motor rehabilitation system with social robotic interfaces that maximizes rehabilitation dosage. The project will provide insights into motor development in special needs children suggesting a new early intervention paradigm for children with motor disabilities.
| Zehfroosh, Ashkan; Kokkoni, Elena; Tanner, Herbert G et al. (2017) Learning models of Human-Robot Interaction from small data. Mediterr Conf Control Automation 2017:223-228 |
