Zhongkai Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Zhongkai Zhang has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Chungang Zhuang
School of Mechanical Engineering, Shanghai Jiao Tong University, China
7 shared publications - Hongbo Hu
School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai, China
7 shared publications - Pengxin Zha
Shanghai Jiao Tong University
4 shared publications Huaiwu Zou
Shanghai Aerospace System Engineering Research Institute, China
1 shared publication- Wenyu Hu
Shanghai Jiao Tong University
1 shared publication - Yan Lu
Robotics Institute, School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai, China
1 shared publication
Latest Publications
- 21 April 2026
- Published by SAGE Publications in Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science
- Vol. 240 (14) , 5355-5366
Nonlinear 6-DOF stiffness modeling and identification for industrial robot
Stiffness modeling of industrial robots is a foundational technology for elucidating deformation mechanisms and enabling deformation compensation. Current research predominantly relies on a 1-degree-of-freedom (1-DOF) joint stiffness assumption, which inadequately characterizes the nonlinear stiffness behavior of industrial robots under complex six-dimensional end-effector loadings. Furthermore, existing 6-DOF stiffness models fail to comprehensively account for the nonlinear coupling effects between the end-effector wrench and configuration variations. To enhance deformation prediction accuracy, this paper proposes a 6-DOF nonlinear stiffness model that precisely characterizes these nonlinear properties by analytically modeling the coupling effects between the robot’s configuration and the end-effector wrench. Based on this model, a composite global optimization index is proposed to enhance parameters identification accuracy through the strategic selection of optimal robot configurations. Experimental results demonstrate that the proposed stiffness model achieves a significant reduction in deformation prediction errors, with the mean positional deformation prediction error reduced by 74% and the mean angular deformation prediction error reduced by 94% compared to an existing 1-DOF stiffness model.
- 21 April 2026
- Published by SAGE Publications in Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science
- Vol. 240 (14) , 5355-5366
Nonlinear 6-DOF stiffness modeling and identification for industrial robot
Stiffness modeling of industrial robots is a foundational technology for elucidating deformation mechanisms and enabling deformation compensation. Current research predominantly relies on a 1-degree-of-freedom (1-DOF) joint stiffness assumption, which inadequately characterizes the nonlinear stiffness behavior of industrial robots under complex six-dimensional end-effector loadings. Furthermore, existing 6-DOF stiffness models fail to comprehensively account for the nonlinear coupling effects between the end-effector wrench and configuration variations. To enhance deformation prediction accuracy, this paper proposes a 6-DOF nonlinear stiffness model that precisely characterizes these nonlinear properties by analytically modeling the coupling effects between the robot’s configuration and the end-effector wrench. Based on this model, a composite global optimization index is proposed to enhance parameters identification accuracy through the strategic selection of optimal robot configurations. Experimental results demonstrate that the proposed stiffness model achieves a significant reduction in deformation prediction errors, with the mean positional deformation prediction error reduced by 74% and the mean angular deformation prediction error reduced by 94% compared to an existing 1-DOF stiffness model.
- 19 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. PP (10834435) , 1-12
Modeling and Compensation of Backlash-Induced Dynamics Error in Industrial Robots With a PINN-Based Approach
Robotic dynamics modeling and identification methods lay the foundation for higher precision applications of robots. However, challenges remain in compensating for nonlinear characteristics such as joint backlash. Specifically, the serial kinematics of multiaxis serial robots amplify the impact of backlash on dynamics. This article addresses the backlash phenomenon in six degrees of freedom (DOF) serial robots and proposes a physics-informed neural network (PINN)-based approach for modeling and compensating dynamics error. First, a temporal detection and classification strategy is designed based on the formation mechanism of backlash. Next, an error model based on Gaussian basis functions is proposed, and a corresponding PINN is constructed to implement online compensation for backlash-induced dynamics error. Finally, the error compensation performance of the proposed method is validated experimentally, achieving average error reductions of 45.93% and 34.19% on two 6-DOF robots respectively, outperforming other advanced methods. Additionally, the designed compensation strategy ensures the superior robustness of the proposed method.
- 19 March 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. PP (10834435) , 1-12
Modeling and Compensation of Backlash-Induced Dynamics Error in Industrial Robots With a PINN-Based Approach
Robotic dynamics modeling and identification methods lay the foundation for higher precision applications of robots. However, challenges remain in compensating for nonlinear characteristics such as joint backlash. Specifically, the serial kinematics of multiaxis serial robots amplify the impact of backlash on dynamics. This article addresses the backlash phenomenon in six degrees of freedom (DOF) serial robots and proposes a physics-informed neural network (PINN)-based approach for modeling and compensating dynamics error. First, a temporal detection and classification strategy is designed based on the formation mechanism of backlash. Next, an error model based on Gaussian basis functions is proposed, and a corresponding PINN is constructed to implement online compensation for backlash-induced dynamics error. Finally, the error compensation performance of the proposed method is validated experimentally, achieving average error reductions of 45.93% and 34.19% on two 6-DOF robots respectively, outperforming other advanced methods. Additionally, the designed compensation strategy ensures the superior robustness of the proposed method.
- 25 October 2025
- Published by Springer Nature
- p. 39-51
Dual-Channel Adaptive Impedance Algorithm with Leveling Module in Dual-Arm Collaborative Robots
- 25 October 2025
- Published by Springer Nature
- p. 39-51
Dual-Channel Adaptive Impedance Algorithm with Leveling Module in Dual-Arm Collaborative Robots
- 25 October 2025
- Published by Springer Nature
- p. 27-38
Dynamics Modeling and Vibration Suppression of Industrial Robots Handling Flexible Payloads
- 25 October 2025
- Published by Springer Nature
- p. 27-38
Dynamics Modeling and Vibration Suppression of Industrial Robots Handling Flexible Payloads
- 2 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (7) , 7619-7626
Unified Gravitational and Elasto-Geometrical Calibration for an Industrial Robot Using Closed-Form Formulation
The absolute positioning accuracy of industrial robots is significant in modern manufacturing scenarios. Mainstream approaches mostly focus on calibrating kinematic and joint compliance models to compensate for position errors. However, previous elasto-geometrical calibrations typically employ a priori gravity parameters (e.g., link mass) to estimate the compliant joint deformation, which inevitably leads to a loss of accuracy. To obtain more accurate results, this paper proposes a unified gravitational and elasto-geometrical calibration method based on the closed-form expression. The proposed method establishes a unified model that integrates kinematic, joint stiffness, and gravity parameters, thereby enabling the handling of inter-model coupling characteristics. An analytic derivative of gravity torque is derived, and an improved adaptive damping Levenberg-Marquardt algorithm is utilized to accelerate calibration convergence. Finally, the position error of a serial industrial robot is reduced from 3.329 mm to 0.251 mm using the proposed method. The load generalization experiments further validate the effectiveness of the proposed method.
- 2 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (7) , 7619-7626
Unified Gravitational and Elasto-Geometrical Calibration for an Industrial Robot Using Closed-Form Formulation
The absolute positioning accuracy of industrial robots is significant in modern manufacturing scenarios. Mainstream approaches mostly focus on calibrating kinematic and joint compliance models to compensate for position errors. However, previous elasto-geometrical calibrations typically employ a priori gravity parameters (e.g., link mass) to estimate the compliant joint deformation, which inevitably leads to a loss of accuracy. To obtain more accurate results, this paper proposes a unified gravitational and elasto-geometrical calibration method based on the closed-form expression. The proposed method establishes a unified model that integrates kinematic, joint stiffness, and gravity parameters, thereby enabling the handling of inter-model coupling characteristics. An analytic derivative of gravity torque is derived, and an improved adaptive damping Levenberg-Marquardt algorithm is utilized to accelerate calibration convergence. Finally, the position error of a serial industrial robot is reduced from 3.329 mm to 0.251 mm using the proposed method. The load generalization experiments further validate the effectiveness of the proposed method.
- 22 January 2025
- Published by Springer Nature
- p. 300-315
A Hybrid Elasto-Geometrical Calibration Method for Industrial Robot Using Only Position Measurement
- 22 January 2025
- Published by Springer Nature
- p. 300-315
A Hybrid Elasto-Geometrical Calibration Method for Industrial Robot Using Only Position Measurement