Distributed prescribed-time leader–follower tracking consensus control for high-order nonlinear MASs

Jiaping Qiang, Li Li, Yuanqing Xia

TL;DR

Researchers have developed a novel control strategy that allows a group of high-order nonlinear robots to synchronize with a leader within a strictly pre-defined time window. Unlike previous methods, this "prescribed-time" approach works regardless of the robots' starting positions and avoids the computational nightmares typically associated with complex nonlinear systems.

What Problem Does This Paper Address?

In the world of Multi-Agent Systems (MASs)—think drone swarms, autonomous underwater vehicles, or robotic assembly lines—coordination is everything. Specifically, "leader-follower tracking consensus" is the goal of getting every agent in a group to follow a leader’s trajectory perfectly.

While we have "finite-time" and "fixed-time" control methods, they have a major flaw: the time it takes to synchronize usually depends on how far apart the agents start or requires tuning complex parameters. If you need a swarm to dock with a moving spacecraft or intercept a target, you need a prescribed-time guarantee—a "hard deadline" that the user can set.

Furthermore, most real-world systems are "high-order nonlinear" systems. Controlling these usually involves a technique called "back-stepping," which often leads to a "differential explosion"—a mathematical mess where you have to calculate increasingly complex derivatives at every step, making it nearly impossible to implement on actual hardware.

Methods

The authors propose a two-stage strategy to solve these issues:

  1. Distributed Prescribed-Time State Observer (DPTSO): Since followers might not always see the leader directly, each agent uses this observer to estimate the leader's state. By using a specialized time-varying rate function, every follower "locks on" to the leader's estimated position within a user-defined time ($T_1$).
  2. Cascade Control with Dynamic Gains: Instead of traditional back-stepping, the team used a cascade control framework. They introduced a dynamic gain ($L_i$) that automatically adjusts to offset the system's nonlinearities. This approach effectively "flattens" the complexity, allowing for a straightforward controller design that avoids the dreaded differential explosion.
  3. Time-Varying Scaling: The core "magic" lies in functions that scale toward infinity as the clock approaches the deadline. This forces the tracking error to zero exactly when the user wants it to, while ensuring the control signals remain smooth and bounded.

Datasets

As a theoretical control paper, this work does not use external datasets. Instead, the authors validated their mathematical proofs using numerical simulations of a group consisting of one leader and four followers. The agents were modeled as second-order nonlinear systems (representing simplified physical dynamics) to demonstrate the controller's performance.

Key Insights and Ideas

  • Initial State Independence: The most striking feature is that the convergence time is independent of the initial conditions. Whether the agents start 1 unit or 50 units away, they all reach consensus at the exact same pre-specified moment.
  • Simplicity by Design: By bypassing the step-by-step virtual control laws of back-stepping, the resulting controller is much simpler to design and adjust. This makes it a strong candidate for practical engineering applications where processor power is limited.
  • Smoothness Matters: The control action is continuous and "almost $C^1$ smooth." In robotics, this is vital because it prevents "chattering"—rapid, jerky movements that can damage motors and actuators.
  • Post-Deadline Stability: The system doesn't just stop or break after the deadline; the designed functions ensure the agents continue to operate and maintain consensus indefinitely after the initial synchronization.

Key Results

The authors tested their method against three different groups of initial starting positions:

  • Precision Timing: In simulations with a total prescribed time of 1.7 seconds ($T_1=0.5s$ for observation, $T_2=1.2s$ for control), the tracking error for all agents hit zero precisely at the 1.7-second mark.
  • Flexibility: When the deadline was tightened to 1.3 seconds ($T_1=0.3s, T_2=1.0s$), the system adapted perfectly, reaching consensus faster without needing to redesign the entire controller.
  • Superiority over Fixed-Time: Compared to existing "fixed-time" algorithms, this method showed higher steady-state accuracy and a much simpler implementation process, though the authors noted it can result in higher initial "overshoot" (a temporary surge in control effort) to meet tight deadlines.

Limitations and Future Work

The primary limitation acknowledged by the authors is that their current model assumes a "clean" environment. In the real world, robots face external disturbances like wind, friction, or sensor noise.

Future Work: The team plans to extend this prescribed-time framework to handle "disturbed" high-order nonlinear systems, ensuring that the hard deadlines can still be met even when the environment is unpredictable.

Summary updated

Abstract

<title>Abstract</title> A novel distributed prescribed-time leader-follower tracking consensus control strategy is proposed for high-order nonlinear multi-agent systems (MASs) with lower triangular time-varying dynamic in directed communication topology. Firstly, a distributed prescribed-time state observer (DPTSO) is presented for follower to observe the leader’s state, transforming the consensus problem into a tracking control problem that the follower tracks the observations of the DPTSO. Then, a distributed prescribed-time controller is developed with the cascade control framework and dynamic control, which avoids differential explosion in traditional back-stepping control. Convergence time of both the DPTSO and the distributed prescribed-time controller is not only explicitly pre-specified but also determined regardless of the initial state of system and control parameters. Finally, it is demonstrated that the closed-loop systems realize prescribed-time full state leader-follower tracking consensus. Simulation results show that the method is effective and feasible.
BibTeX
@inproceedings{qiang2023distributed,
  author = {Jiaping Qiang and Li Li and Yuanqing Xia},
  title = {{Distributed prescribed-time leader–follower tracking consensus control for high-order nonlinear MASs}},
  booktitle = {Nonlinear},
  year = {2023},
  doi = {10.1007/s11071-023-09051-z},
}

Loading related content...

Figure