Efficient hydraulic bipedal robot control with model predictive control

Abstract

A hydraulic power unit (HPU) is a essential device to supply pressure and flow rate in a hydraulic drive system. The existing HPU have disadvantages in that they are relatively large and heavy, and its driving method is also inefficient in terms of the energy-efficiency. In this work, we will present a light-weight HPU design method for embedding it on a biped robot to perform various motions and an energy efficient control method to prevent overheating of the oil and to extend operating time of the robot. The HPU introduced in this work has a structure in which an electromagnet motor and a pump are integrated into a single module, and the module is immersed in hydraulic oil. This layout is advantageous not only for a miniaturized design, but also for improving the heat problem that occurs when delivering highly pressurized energy. A relationship between motor geometric parameters and the thermodynamic behavior of the motor will be expressed mathematically, and a process to design the shape of the motor for stable operation under high-pressure operation will be provided. Also, we will present a control method that can reduce the energy consumed by establishing a energy loss model occurring throughout the robot. This control method, which is based on model predictive control (MPC), provides an optimized pump speed minimizing the energy loss caused in the overall system while supplying enough pressure to achieve various tasks with the robot. Finally, we will verify the functionality and effectiveness of the HPU and its controller by applying the proposed methodology to LIGHT, a hydraulic biped robot.