A Single Parameter-Based Adaptive Approach to Robotic Manipulators With Finite Time Convergence and Actuator Fault

Shuang Zhang, Pengxin Yang, Linghuan Kong, Guang Li, Wei HeIEEE Access 2020 · DOI

Abstract

This paper presents a novel single-parameter adaptive finite-time fault-tolerant control (FTC) scheme designed for n-link robotic systems. The proposed method addresses the simultaneous challenges of actuator faults, external disturbances, system parameter uncertainties, and physical saturation constraints. By integrating a nonsingular fast terminal sliding mode (NFTSM) control with a unique single-parameter adaptive law, the system achieves robust tracking and guaranteed finite-time convergence. Unlike traditional methods that require complex estimation of multiple variables, this approach "lumps" uncertainties into a single adaptive parameter, significantly reducing computational overhead. Simulation results on a 2-DOF planar manipulator demonstrate the effectiveness and efficiency of the proposed active FTC (AFTC) strategy compared to passive alternatives.

Introduction

In modern industrial and service robotics, the reliability of robotic manipulators is paramount. However, real-world operations are rarely ideal. Robotic systems frequently encounter three primary obstacles: model uncertainties (such as unknown mass or friction), external disturbances (environmental forces), and actuator faults (mechanical wear or power loss). Furthermore, all physical motors are subject to saturation constraints, meaning they have a finite limit on the torque they can provide.

Traditional control strategies often struggle with these issues simultaneously. Many rely on asymptotic convergence, where the error reaches zero only as time approaches infinity. Others utilize complex observers or neural networks to estimate every unknown variable, which demands high computational power and can introduce latency in real-time applications. This paper proposes a streamlined Active Fault Tolerant Control (AFTC) that simplifies the estimation process while guaranteeing that the robot reaches its target within a specific, finite time window.

Methodology

1. Nonsingular Fast Terminal Sliding Mode (NFTSM)

The foundation of the control scheme is the NFTSM. Sliding mode control is a robust technique that forces the system's state onto a predefined mathematical surface where errors are eliminated. The "Nonsingular" aspect ensures that the control law avoids mathematical singularities (such as division by zero), while the "Fast Terminal" component ensures that the system state converges to the equilibrium point in finite time, rather than just approaching it asymptotically.

2. The Single-Parameter Adaptive Law

The core innovation of this research is the simplification of the uncertainty estimation. Instead of attempting to calculate the exact value of every individual disturbance or fault, the authors define a single upper bound for the total system error. A single adaptive parameter, denoted as $\hat{\Lambda}$, is used to estimate this lumped uncertainty. This reduction from multiple variables to one significantly decreases the time required for online calculations, making the controller ideal for low-power embedded processors.

3. Active Fault Tolerant Control (AFTC)

The system employs an "Active" strategy. While passive controllers are designed to be generally robust, they often sacrifice performance during normal operation. The proposed AFTC uses the single adaptive parameter to actively sense changes in the system's health—such as a sudden loss of motor torque—and adjusts the control torque in real-time to compensate for the specific fault detected.

Experimental Validation

The methodology was validated using a simulated 2-degree-of-freedom (2-DOF) planar robotic manipulator. The simulation modeled standard physical parameters (mass, length, and inertia) and introduced rigorous testing conditions:

  • Actuator Faults: A simulated 90% loss of torque effectiveness in the actuators.
  • Disturbances: External time-varying forces applied to the joints.
  • Saturation: Hard limits on the maximum torque the motors could output.

Key Results and Discussion

The simulation results highlighted several critical advantages of the single-parameter AFTC:

  • Rapid Fault Recovery: When the actuators suffered a 90% power loss at $t=3s$, the AFTC successfully adjusted the control parameters within 0.3 seconds of the fault's onset, maintaining the desired trajectory.
  • Superiority over Passive Control: In direct comparison with a Passive Fault Tolerant Controller (PFTC), the proposed AFTC exhibited significantly lower tracking errors and a faster recovery rate during fault periods.
  • Finite-Time Convergence: The mathematical guarantee of convergence within a calculated time window ($t_1 + t_2$) was confirmed, providing a level of predictability essential for high-precision robotic tasks.
  • Computational Efficiency: By utilizing only a single adaptive parameter, the controller maintained stability and performance without the heavy processing load typical of multi-parameter adaptive systems.

Conclusion

This paper demonstrates that complex robotic control problems do not always require complex solutions. By lumping system uncertainties into a single adaptive parameter and utilizing NFTSM, the researchers developed a robust, finite-time controller capable of handling severe actuator faults and physical constraints. While currently validated in simulation, this approach offers a promising path toward more efficient and resilient industrial robots. Future work will focus on implementing this controller on physical hardware and exploring its application in more unpredictable environments, such as underwater or aerial robotics.

Summary updated

BibTeX
@article{zhang2020single,
  author = {Shuang Zhang and Pengxin Yang and Linghuan Kong and Guang Li and Wei He},
  title = {{A Single Parameter-Based Adaptive Approach to Robotic Manipulators With Finite Time Convergence and Actuator Fault}},
  journal = {IEEE Access},
  year = {2020},
  doi = {10.1109/access.2020.2966639},
}
Figure