Robot arms require actuators that are powerful, precise and safe. The safety concern is amplified when these robots work closely with people in collaborative applications. This thesis investigates the design and implementation of hybrid pneumatic-electric actuators (HPEA) for use in robot arms, particularly those intended for collaborative applications. The initial focus was on improving the control of an existing single HPEA-driven rotary joint. The torque is produced by four pneumatic cylinders connected in parallel with a small DC motor. The DC motor is directly connected to the output shaft. A cascaded control system is designed that consists of an outer position control loop and an inner pressure control loop. The pressure controller is based on a novel inverse valve model. High precision position tracking control is achieved due to the combination of the model-based pressure controller, model-based position controller, adaptive friction compensator and offline payload estimator. Experiments are performed with the actuator prototype rotating a link and payload with a rotational inertia equivalent to a linear actuator moving a 573 kg mass. Averaged over five tests, a root-mean-square error of 0.024° and a steady-state error (SSE) of 0.0045° are achieved for a fast multi-cycloidal trajectory. This SSE is almost ten times smaller than the best value reported for previous HPEAs. An offline payload estimation algorithm is used to improve the control system’s robustness. The superior safety of the HPEA is shown by modeling and simulating a constrained robot-head impact, and comparing the result with equivalent electric and pneumatic actuators. This research produced two journal papers.
Since HPEAs are redundant actuators that combine the large force, low bandwidth characteristics of pneumatic actuators with the large bandwidth, small force characteristics of electric actuators, the effect of using optimization-based input allocation for HPEAs was studied. The goal was to improve the HPEA’s performance by distributing the required input (force or torque) between the redundant actuators in accordance with each actuator’s advantages and limitations. Three novel model-predictive control (MPC) approaches are designed to solve the position tracking and input allocation problems using convex optimization. The approaches are simulated on a HPEA-driven system and compared to a conventional linear controller without active input allocation. The first MPC approach uses a model that includes the dynamics of the payload and pneumatics; and performs the motion control using a single loop. The latter methods simplify the MPC law by separating the position and pressure controllers. Although the linear controller is the most computationally efficient, it is inferior to the MPC-based controllers in position tracking and force allocation performance. The third MPC-based controller design demonstrated the best position tracking with root mean square errors of 46%, 20%, and 55% smaller than the other three approaches. It also demonstrated sufficient speed for real-time operation. This research produced one journal paper.
The research continued with the design and implementation of a two degree-of-freedom HPEA-driven arm. A HPEA-driven “elbow” joint is designed and added to the existing “shoulder” joint. The force from a single pneumatic cylinder is converted into torque using a 4-bar linkage. To eliminate backlash and keep the weight of the arm low, a 2nd smaller DC motor is directly connected to the joint. The kinematic and kinetic models of the new arm, as well as the geometry of the new elbow joint are studied. The resulting joint design is implemented, tested and controlled. This joint could achieve a SSE of 0.0045° in spite of its nonlinear joint geometry. The arm is experimentally tested for simultaneous tracking control of the two joints, and for end-effector position tracking in Cartesian space. The end-effector is able to follow a circular trajectory in pneumatic mode with position errors below 0.005 m.