Safety control of monotone systems with bounded uncertainties

Sadra Sadraddini, Călin Belta

Abstract

Monotone systems are prevalent in models of engineering applications such as transportation and biological networks. In this paper, we investigate the problem of finding a control strategy for a discrete time positive monotone system with bounded uncertainties such that the evolution of the system is guaranteed to be confined to a safe set in the state space for all times. By exploiting monotonicity, we propose an approach to this problem which is based on constraint programming. We find control strategies that are based on repetitions of finite sequences of control actions. We show that, under assumptions made in the paper, safety control of monotone systems does not require state measurement. We demonstrate the results on a signalized urban traffic network, where the safety objective is to keep the traffic flow free of congestion.

Introduction

Ensuring safety in complex networks—such as traffic grids or biological pathways—is a fundamental challenge in control theory. Safety is defined as preventing a system from entering a forbidden state, like a traffic jam or chemical overflow. Traditionally, this requires finding a Robust Control Invariant Set (RCIS), a region where control actions can always counteract external disturbances. However, calculating an RCIS for high-dimensional systems is computationally expensive and often requires continuous state feedback.

This paper introduces a computationally efficient method for ensuring safety in positive monotone systems. By leveraging monotonicity, the authors demonstrate that repeating a finite sequence of control actions (an "s-sequence") can guarantee indefinite safety without requiring real-time state measurements or feedback.

Methodology: Monotonicity and Lower-Sets

The core of the approach lies in the property of monotonicity: if a system starts in a "smaller" state (e.g., fewer vehicles), future states will remain "smaller" than those originating from a "larger" state, given identical control inputs.

A lower-set $S \subset \mathbb{R}^2$ representing a safe region.

The authors define the safe region as a lower-set. Intuitively, if a certain state is safe, any state "smaller" than it is also safe. This assumption simplifies the mathematical search for control policies. The search for "s-sequences"—finite control sequences that return the system to a safe state from the "worst-case" boundary—is formulated as a Mixed-Integer Linear Programming (MILP) or Satisfiability Modulo Theories (SMT) problem. This allows the use of powerful off-the-shelf solvers to find open-loop control strategies that do not require sensors.

Experimental Evaluation

The method was evaluated on simulated benchmarks, most notably an urban traffic network consisting of 12 links and 6 junctions.

Urban traffic network with 12 links and 6 junctions used for evaluation.

In the traffic network model, the goal is to prevent congestion despite exogenous vehicle arrivals. The MILP solver successfully identified a valid safety strategy for this network in just 79 seconds. Traditional discretization methods would be intractable for a system of this complexity.

Key Results and Insights

  • Open-Loop Safety: For monotone systems, a fixed "rhythm" of control actions (like a pre-timed traffic light cycle) can mathematically guarantee safety, eliminating the need for constant sensing.
  • Computational Efficiency: The use of constraint programming bypasses the "curse of dimensionality" typical of geometric RCIS calculations.
  • Robustness: The paper provides a theoretical "almost necessity" result, suggesting that if a short s-sequence cannot be found, a robust safety controller likely does not exist or is extremely fragile.

Conclusion and Future Work

While highly effective for systems where safety is defined by upper bounds (lower-sets), the method is currently limited to such structures. It would not apply to systems requiring a specific operating range (e.g., temperature maintenance). Future work aims to tighten the theoretical bounds of the "almost necessity" proof and explore larger, more complex safe set geometries.

Summary updated

Abstract

Monotone systems are prevalent in models of engineering applications such as transportation and biological networks. In this paper, we investigate the problem of finding a control strategy for a discrete time positive monotone system with bounded uncertainties such that the evolution of the system is guaranteed to be confined to a safe set in the state space for all times. By exploiting monotonicity, we propose an approach to this problem which is based on constraint programming. We find control strategies that are based on repetitions of finite sequences of control actions. We show that, under assumptions made in the paper, safety control of monotone systems does not require state measurement. We demonstrate the results on a signalized urban traffic network, where the safety objective is to keep the traffic flow free of congestion.
BibTeX
@misc{sadraddini2016safety,
  author = {Sadra Sadraddini and Călin Belta},
  title = {{Safety control of monotone systems with bounded uncertainties}},
  year = {2016},
  eprint = {1603.07419},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1603.07419},
}
Figure