An optimized design of a parallel robot for gait training

The guidelines for enhancing robot-assisted training for post-stroke survivors head towards increasing exercise realism and variability; in particular lower limb rehabilitation needs the patient to feel challenged to adapt his locomotion and dynamic balance capabilities to different virtual ground scenarios. This paper proposes a design for a robot whose end-effector acts as a footplate to be in permanent contact with the user's foot during practice: the structure is such that it enables the user's foot to rotate around three axis, differently from what is currently available in the research for gait training; the parallel kinematic structure and the dimensional synthesis allow a suitable range of motion and aim at limiting device mass, footprint and reaction forces on the actuators when rendering virtual ground. The employed methodology has been validated using ground reaction forces data relative to stroke survivors.