We designed a training robot for home-based limb rehabilitation by integrating joint-automated-combat-knowledge (JACK) virtual simulation. The robot addresses the need for motor function rehabilitation of hemiplegic patients at home. To design the robot, we conducted a literature review on upper limb hemiplegia and conventional rehabilitation robots, and analyzed the training method for patients with upper limb hemiplegia using the rapid upper limb assessment method and JACK software. As a result, posture risks associated with various movements without robotic assistance were assessed and identified. The needs of the patients were also identified through interviews to refine the design of the upper limb rehabilitation robot. The applicability of the robot was evaluated by JACK simulation analysis, and the results demonstrated that the robot significantly enhanced patient comfort and safety in rehabilitation training.