This paper presents a novel dynamics-based human–robot collaboration (HRC) control method with a remote center-of-motion (RCM) constraint. The existing works rely on prescribed main task trajectories and regard the RCM constraint as a secondary task, making them inapplicable in the fully interactive mode under HRC. Our work imposes a virtual RCM constraint on the interactive HRC process so that the robot’s motion conforms to human intentions while keeping the robot’s end-effector shaft always passing through a fixed (RCM) point. In our approach, the task coordinates of the RCM constraint and its Jacobian matrix are formulated, and a task control law with a computed torque controller is proposed to guarantee the convergence of the RCM error. In the null space of the RCM constraint, a mass-damping impedance control law is used to make the robot motion conform to human interactions. To address the uncertainties of both the dynamic model and external interactions of the robot, a nonlinear disturbance observer is employed to estimate the lumped disturbance projected to the task space of the RCM for steady error elimination. We also show that the robot RCM task approaches a singularity as the RCM error approaches zero. A least-squares damping inversion method is used to map the task-space motion to the joint space near the singularity. Experiments are performed to validate the effectiveness of our method, and the results show that the maximum RCM error is less than 0.85 mm during fast HRC interactions and converges to zero when the interactions cease.

中文翻译：

---

具有远程运动中心约束的动态控制，用于人机协作

本文提出了一种具有远程运动中心 （RCM） 约束的新型基于动力学的人机协作 （HRC） 控制方法。现有工作依赖于规定的主要任务轨迹，并将 RCM 约束视为次要任务，使其不适用于 HRC 下的完全交互模式。我们的工作对交互式 HRC 过程施加了虚拟 RCM 约束，以便机器人的运动符合人类的意图，同时保持机器人的末端执行器轴始终通过固定 （RCM） 点。该方法给出了 RCM 约束的任务坐标及其雅可比矩阵，并提出了一个带有计算扭矩控制器的任务控制律来保证 RCM 误差的收敛性。在 RCM 约束的零空间中，使用质量阻尼阻抗控制定律使机器人运动符合人类交互。为了解决机器人动力学模型和外部交互的不确定性，采用非线性干扰观测器来估计投射到 RCM 任务空间的集总干扰，以稳定地消除误差。我们还表明，当 RCM 误差接近零时，机器人 RCM 任务接近奇点。使用最小二乘阻尼反演方法将任务空间运动映射到奇点附近的关节空间。通过实验验证了我们方法的有效性，结果表明，在快速 HRC 交互期间，最大 RCM 误差小于 0.85 mm，当交互停止时收敛为零。