Model Predictive Fuzzy Control: A Hierarchical Multi-Agent Control Architecture for Outdoor Search-and-Rescue Robots

TL;DR

This paper tackles autonomous SaR in unknown outdoor environments by proposing Model Predictive Fuzzy Control (MPFC), a two-layer hierarchical framework that combines local fuzzy logic controllers (FLC) for fast, heuristic path planning with a centralised model predictive control (MPC) layer that tunes FLC parameters over a finite horizon. The architecture enables real-time decision making at the robot level while injecting global optimization through MPC, offering performance that approaches centralised MPC at a fraction of the computational cost. The authors develop a grid-based 2D disaster model with fire dynamics, wind effects, debris, and victim probabilities to evaluate MPFC against centralised MPC and a pre-tuned FLC baseline across various scenarios, robot counts, and configurations. Results indicate MPFC generally improves performance and robustness with reduced computation, and local maps or decentralised MPFC can further enhance scalability for larger, dynamic SaR environments, making the approach practically attractive for real-time multi-robot missions.

Abstract

Autonomous robots deployed in unknown search-and-rescue (SaR) environments can significantly improve the efficiency of the mission by assisting in fast localisation and rescue of the trapped victims. We propose a novel integrated hierarchical control architecture, called model predictive fuzzy control (MPFC), for autonomous mission planning of multi-robot SaR systems that should efficiently map an unknown environment: We combine model predictive control (MPC) and fuzzy logic control (FLC), where the robots are locally controlled by computationally efficient FLC controllers, and the parameters of these local controllers are tuned via a centralised MPC controller, in a regular or event-triggered manner. The proposed architecture provides three main advantages: (1) The control decisions are made by the FLC controllers, thus the real-time computation time is affordable. (2) The centralised MPC controller optimises the performance criteria with a global and predictive vision of the system dynamics, and updates the parameters of the FLC controllers accordingly. (3) FLC controllers are heuristic by nature and thus do not take into account optimality in their decisions, while the tuned parameters via the MPC controller can indirectly incorporate some level of optimality in local decisions of the robots. A simulation environment for victim detection in a disaster environment was designed in MATLAB using discrete, 2-D grid-based models. While being comparable from the point of computational efficiency, the integrated MPFC architecture improves the performance of the multi-robot SaR system compared to decentralised FLC controllers. Moreover, the performance of MPFC is comparable to the performance of centralised MPC for path planning of SaR robots, whereas MPFC requires significantly less computational resources, since the number of the optimisation variables in the control problem are reduced.

Figures (50)

• Figure 1: Different time scales used for modelling the SaR environment: All local simulation time steps starting from $k^{\textrm{rob}}$ that fall within the global simulation time steps $k$ and $k+1$ are shown by set $\mathbb{K}_k$.
• Figure 2: Illustration of a SaR environment given via a 2D satellite image (a) and modelled by 2D grids of cells (b), and 2D representation of an environment that has been discretised via cells of different sizes (c).
• Figure 3: All the $(l, q)$ neighbouring cells of cell $(i, j)$ from which the fire can spread to $(i, j)$, based on the wind speed: dotted cells for $m^{\textrm{velocity}}_{lq}(k) < 1~\frac{\textrm{m}}{\textrm{s}}$; dotted and lined cells for $1~\frac{\textrm{m}}{\textrm{s}} \leq m^{\textrm{velocity}}_{lq}(k) \leq 5~\frac{\textrm{m}}{\textrm{s}}$; dotted, lined and solid cells for $m^{\textrm{velocity}}_{lq}(k) > 5~\frac{\textrm{m}}{\textrm{s}}$.
• Figure 4: Corrected heading angle for achieving an air velocity for the flying robot that, despite the wind flow, allows the robot to fly in the desired direction (aligned with the ground velocity).
• Figure 5: MPFC architecture at time step $k^{\textrm{ctrl}}$, with two layers specified by dashed rectangles: The fuzzy inference systems control the multi-agent system directly, in a decentralised way. The model predictive control layers tunes, with a frequency that is generally lower than the control frequency, the parameters of the local fuzzy inference systems in order to improve the global performance of the multi-agent system.