Admittance-Based Controller Design for Physical Human–Robot Interaction in the Constrained Task Space
Wei He,Chengqian Xue,Xinbo Yu,Zhijun Li,Chenguang Yang
DOI: https://doi.org/10.1109/tase.2020.2983225
IF: 6.636
2020-01-01
IEEE Transactions on Automation Science and Engineering
Abstract:In this article, an admittance-based controller for physical human-robot interaction (pHRI) is presented to perform the coordinated operation in the constrained task space. An admittance model and a soft saturation function are employed to generate a differentiable reference trajectory to ensure that the end-effector motion of the manipulator complies with the human operation and avoids collision with surroundings. Then, an adaptive neural network (NN) controller involving integral barrier Lyapunov function (IBLF) is designed to deal with tracking issues. Meanwhile, the controller can guarantee the end-effector of the manipulator limited in the constrained task space. A learning method based on the radial basis function NN (RBFNN) is involved in controller design to compensate for the dynamic uncertainties and improve tracking performance. The IBLF method is provided to prevent violations of the constrained task space. We prove that all states of the closed-loop system are semiglobally uniformly ultimately bounded (SGUUB) by utilizing the Lyapunov stability principles. At last, the effectiveness of the proposed algorithm is verified on a Baxter robot experiment platform. Note to Practitioners-This work is motivated by the neglect of safety in existing controller design in physical human-robot interaction (pHRI), which exists in industry and services, such as assembly and medical care. It is considerably required in the controller design for rigorously handling constraints. Therefore, in this article, we propose a novel admittance-based human-robot interaction controller. The developed controller has the following functionalities: 1) ensuring reference trajectory remaining in the constrained task space: a differentiable reference trajectory is shaped by the desired admittance model and a soft saturation function; 2) solving uncertainties of robotic dynamics: a learning approach based on radial basis function neural network (RBFNN) is involved in controller design; and 3) ensuring the end-effector of the manipulator remaining in the constrained task space: different from other barrier Lyapunov function (BLF), integral BLF (IBLF) is proposed to constrain system output directly rather than tracking error, which may be more convenient for controller designers. The controller can be potentially applied in many areas. First, it can be used in the rehabilitation robot to avoid injuring the patient by limiting the motion. Second, it can ensure the end-effector of the industrial manipulator in a prescribed task region. In some industrial tasks, dangerous or damageable tools are mounted on the end-effector, and it will hurt humans and bring damage to the robot when the end-effector is out of the prescribed task region. Third, it may bring a new idea to the designed controller for avoiding collisions in pHRI when collisions occur in the prescribed trajectory of end-effector.