Zhuo Wang

To improve the accuracy of torque estimation and compliance control of the force sensorless, we propose a torque fusion method based on extended Kalman filter (EKF), both the data of motor current and the harmonic reducer torsional deformation are involved. First, a nonlinear EKF is designed based on the motor side dynamic model and joint deformation nonlinear model. The experiment shows that the method overcomes the interference caused by the environment and nonlinear system factors. Then, an active compliance controller is designed with a nested loop framework based on the fused torque. The inner loop computes the dynamic torque as feedforward to compensate for the system's dynamic uncertainty. The outer loop is admittance control to realize the manipulator's active compliance with the fused torque. Experiments on the robotic manipulator show that the proposed torque estimation scheme can reduce the root mean square error (RMSE) to 0.1278N $\cdot$ m and the max error of the joint estimated torque to 0.34N $\cdot$ m. In addition, the force tracking accuracy of the proposed compliant controller can reach 3.6 N, and can be extended to more redundant joints.

Joint stiffness of an elastic-joint robot can be changed according to joint stiffness requirements. A series elastic actuator (SEA) can reduce the contact stiffness between the body and the environment or human, which can further ensure interactive operation in a human–machine-compatible environment. However, the introduction of the SEA improves the complexity of the robot dynamics model. In this paper, we propose a control schema based on link-side acceleration measurement to eliminate the overshoot and vibration in the transient process of force control. An extended Kalman filter (EKF) algorithm that fuses photoelectric encoders and accelerometers is first presented based on the link-side acceleration measurement. Following this, based on the external torque estimation, the vibration reduction control algorithm is designed. The simulation model is built, and the algorithm design and simulation of position control and force control are carried out and finally tested on the real robot platform. The effectiveness of the control algorithm is proved. The experimental results show that the dynamic response of the external force estimation is about 2 ms faster than that of the force sensor, and the error between the estimated external torque and the real external torque is within ±0.16 N·m.