Multirobot-Guided Crowd Evacuation: Two-Scale Modeling and Control

Tongjia Zheng, Zhenyuan Yuan, Mollik Nayyar, Alan R. Wagner, Minghui Zhu, Hai Lin

TL;DR

This paper presents a novel two-scale modeling and control framework for multi-robot-guided crowd evacuation. By combining microscopic human behavior models with macroscopic crowd dynamics, the system uses a small fleet of robots to dynamically guide large crowds to safety, even in the presence of unknown obstacles, and proves the stability of the proposed approach.

What Problem Does This Paper Address?

Emergency evacuations are inherently chaotic and time-critical. While mobile robots offer a promising solution for guiding people to safety, effectively controlling a large number of humans with a limited number of robots is a significant challenge. The core problem is how to leverage micro-level human-robot interactions to indirectly influence the collective behavior of a crowd that vastly outnumbers the robots, ensuring an efficient and safe evacuation. Existing approaches often struggle with the high dimensionality of microscopic models for large crowds or lack the ability to incorporate robot guidance into macroscopic models.

Methods

The paper introduces a two-scale modeling strategy that bridges the gap between microscopic and macroscopic crowd dynamics.

  • Microscopic Modeling: Human behavior is modeled using the widely accepted social force model, which captures interactions between humans, with the environment (including unknown dynamic obstacles), and with robots. Robots are modeled as second-order integrators.
  • Macroscopic Modeling: From the microscopic models, a hydrodynamic model is derived. This model describes the temporal and spatial evolution of crowd density and flow velocity using partial differential equations. This macroscopic view is crucial for understanding and controlling the collective behavior of the crowd.
  • Robot Control Design: The control strategy for the robots is divided into two parts:
  • Position Control: Robots are tasked with exploring and covering the environment using a potential-field-based approach, ensuring they can sense and interact with the crowd.
  • Direction Control: This is the core innovation. Robots dynamically adjust their "navigation force fields" (visualized as directional arrows) based on real-time macroscopic crowd states (density and velocity). This allows robots to indirectly steer the crowd towards safe locations. An adaptive component is included to handle unknown environmental dynamics.
  • Stability Analysis: The authors provide theoretical proofs for the stability of the proposed evacuation algorithm, ensuring that the crowd density converges to a desired state.

Datasets

Not discussed in the paper. The paper focuses on a simulation-based validation of the proposed algorithms.

Key Insights and Ideas

  • Two-Scale Modeling for Indirect Control: The key insight is that by modeling crowd dynamics at both micro and macro levels, robots can exert indirect control. The robots' local interactions (micro) influence the global crowd behavior (macro), which in turn feeds back to influence individual robot control decisions.
  • Dynamic Navigation Force Fields: Robots don't just push people; they create dynamic "guidance fields" that humans naturally follow. The ability to adapt these fields based on crowd density and flow is crucial for efficient evacuation.
  • Coverage vs. Environment Coverage: The stability analysis suggests that robots only need to cover the crowd's current location (Ωρ(t)), not necessarily the entire environment (Ω), to ensure successful evacuation. This is a significant practical advantage, as it means fewer robots might be needed if they can effectively track and guide the moving crowd.
  • Adaptive Control for Unknown Dynamics: The inclusion of an adaptive control law to compensate for unknown forces (like dynamic obstacles) makes the system more robust and practical for real-world scenarios.

Key Results

  • Stability Guarantees: The paper provides theoretical proofs for the uniform ultimate boundedness of the crowd density, velocity, and robot control errors, indicating the algorithm's stability.
  • Exponential Convergence: Under the assumption that unknown dynamics vanish, the crowd density is shown to converge exponentially to the target density.
  • Simulation Validation: Extensive simulations were conducted with varying numbers of humans (50-250) and robots (4-16), and in environments with no obstacles, static obstacles, and dynamic obstacles.
  • The simulations demonstrated that robots effectively explored and covered the environment.
  • The dynamic navigation force fields successfully guided the crowd towards the target safe location.
  • The estimated macroscopic states (density and velocity) closely matched the microscopic simulations.
  • Evacuation rates generally increased with more humans (due to better macroscopic model approximation) and more robots (due to better coverage).
  • Dynamic obstacles reduced evacuation rates, but this effect became negligible with a sufficient number of robots and humans.

Limitations and Future Work

  • Centralized Communication: The current implementation relies on a centralized communication center to collect global crowd information and distribute commands to robots. This is a limitation for decentralized operation.
  • Unstructured Environments: The current framework is primarily tested in structured environments. Extending it to more complex, unstructured settings is a future direction.
  • Robot Number Determination: While simulations explore various robot numbers, a rigorous method for determining the optimal number of robots for a given scenario is not explicitly provided.
  • More Complex Robot Controllers: The paper suggests that more complicated position controllers might be needed for scenarios where the target density is concentrated in a small region, requiring robots to first spread out and then converge.
  • Real-World Deployment: The work is currently validated through simulations. Future work will involve real-world experiments to further test and refine the algorithms.

Summary updated

Abstract

Emergency evacuation describes a complex situation involving time-critical decision-making by evacuees. Mobile robots are being actively explored as a potential solution to provide timely guidance. This work studies a robot-guided crowd evacuation problem where a small group of robots is used to guide a large human crowd to safe locations. The challenge lies in how to use microlevel human-robot interactions to indirectly influence a population that significantly outnumbers the robots to achieve the collective evacuation objective. To address the challenge, we follow a two-scale modeling strategy and explore hydrodynamic models, which consist of a family of microscopic social force models that describe how human movements are locally affected by other humans, the environment, and robots, and associated macroscopic equations for the temporal and spatial evolution of the crowd density and flow velocity. We design controllers for the robots, such that they not only automatically explore the environment (with unknown dynamic obstacles) to cover it as much as possible, but also dynamically adjust the directions of their local navigation force fields based on the real-time macrostates of the crowd to guide the crowd to a safe location. We prove the stability of the proposed evacuation algorithm and conduct extensive simulations to investigate the performance of the algorithm with different combinations of human numbers, robot numbers, and obstacle settings.
BibTeX
@article{zheng2024multirobot,
  author = {Tongjia Zheng and Zhenyuan Yuan and Mollik Nayyar and Alan R. Wagner and Minghui Zhu and Hai Lin},
  title = {{Multirobot-Guided Crowd Evacuation: Two-Scale Modeling and Control}},
  journal = {IEEE Trans. Control. Syst. Technol.},
  year = {2024},
  url = {https://openreview.net/forum?id=9nxnxSifQs},
}
Figure