This article presents the topology and material optimization of an underactuated lower limb robot for gait rehabilitation of stroke survivors. Robot aided lower limb rehabilitation has been studied for the last three decades and a few over-actuated and underactuated robot designs have been proposed in the literature. Over-actuated robots give more controlled motions, whereas underactuated designs allow unconstrained naturalistic motions. Designing an underactuated robot is difficult as it must be lightweight and yet strong enough to scaffold human lower limbs during gait. In this research, a Stephenson III six-bar linkage is modified to be used as single-actuated lower limb robot. The proposed robot design couples two four-bar linkages in a driver-driven mode to provide motions, which are equivalent to the motions from a six-bar linkage. A digital twin of the underactuated robot is developed to conduct in-silico experiments and evaluate the use of three different materials namely, aluminum alloy, structural steel, and fiber reinforced carbon composite. Finite element analysis (FEA) modeling is carried out using the Ansys workbench to assess mechanical performance indices. In order to minimize the overall weight of the robot, structural optimization is carried out using a multi-mode single objective genetic algorithm. Functional and nonfunctional design requirements are formulated as constraints to be used during the optimization experiments. The topology and material optimization presented in this article provides insights into the robot design requirements and the optimization process. As a result of this process, a significant reduction in the robot weight is achieved without compromising the mechanical performance.