Control of a powered, gravity-balanced orthosis for children with limited upper limb strength
Ragonesi, Daniel David
University of Delaware
There are several pediatric musculoskeletal diseases, such as muscular dystrophy, spinal muscular atrophy, and arthrogryposis, characterized by upper limb weakness with minimal or abnormal motor control and sensation. Often a person with one of these conditions is not able to be independent and requires assistance to perform activities of daily living. Specifically, among other gross motor challenges, these patients often are too weak to overcome the weight of their own arm for daily tasks including self-feeding. There have been several engineering devices designed to increase patients' independence through decreasing the power required to perform upper extremity (UE) tasks. One example is the WREX (Wilmington Robotic Exoskeleton), a currently commercialized passive, pediatric, upper-limb orthosis designed to assist children with upper limb weakness. The WREX has an anthropomorphic configuration and uses parallel mechanisms with zero rest-length springs for gravity balancing. Due to limitations including 1) lack of countering force to allow the child to pick up objects of significant weight and 2) difficulty for the children to raise their arm above their head due to misalignment of joints between user and device and/or possible increasing device joint stiffness with shoulder elevation, it was proposed to add motors and a controller to the WREX to increase UE function. Series Elastic Actuation (SEA) was proposed to maintain a compliant interaction between the device and user. However, testing indicated an instability problem, requiring rigid fixation of the motors to the device instead of SEA. After the motors were attached, the problem of human intention needed to be solved. To do this, a 6-axis force sensor was installed at the forearm between the WREX and the user to detect user intention. A force sensor was used to make the control intuitive for the user. Results indicated that an individual static model for each user was needed to extract the user intention from the sensed force. Three different control laws were implemented - (i) friction compensation, as a baseline, (ii) sensed force proportional to control torque; (iii) sensed force proportional to control velocity. Preliminary testing showed that the control laws were able to detect the direction of intended motion, but further, larger studies are merited to determine any superiority of the control laws. Future work should also include testing more sophisticated control laws, and miniaturizing of the device in order to make it feasible for home use.