This paper focuses on robot control problems where energy regeneration is an explicit consideration, and it proposes a methodology for modeling and control design of regenerative motion control systems. The generic model consists of a robotic manipulator where some joints are actively controlled and the remaining joints are energetically self-contained and semi-actively controlled. The model can capture various electromechanical and hydraulic actuator configurations for industrial robots and powered human-assist devices. The basic control approach consists of three steps. First, a virtual control design is conducted by any suitable means. Then, virtual control inputs are enacted by a matching law for the adjustable parameters of the semi-active joints. Finally, the energy storage dynamics are adjusted using design parameters and an optional outer supervisory loop. The method has several attractive features: design simplicity, amenability to simultaneous plant and control design optimization, explicit treatment of energy regeneration, and applicability to multiple domains. This paper emphasizes electromechanical robots whose semi-active joints use ultracapacitors as the single energy storage medium. An internal energy balance equation and associated ideal self-powered operation (ISPO) condition are given for the semi-active joints. This condition is a structural characteristic of the system and independent of the control law. Extensions to handle higher-order dynamics are presented. Also, it is shown that discrepancies between virtual and actual controls can be mapped to parametric uncertainty in the virtual design. Experimental results confirm the validity of the approach.