Information Update Mechanism and Dynamic Path Re-planning in Crowd Evacuation

Information Update Mechanism and Dynamic Path Re-planning in Crowd Evacuation

Problem Description

During emergency evacuations, environmental information may change dynamically (e.g., fire spread, exit blockage, new hazard emergence). Individuals need to adjust their route choices based on real-time information. The core issues of information update mechanisms and dynamic path re-planning include:

  1. How is information transmitted?: Through a central system (e.g., broadcast, electronic displays) or distributed interaction (e.g., communication between individuals)?
  2. Information update frequency and delay: High-frequency updates may overload the system, while low-frequency updates may lead to information lag.
  3. Re-planning trigger conditions: Based on individual location, global events (e.g., exit closure), or local congestion?
  4. Path re-planning algorithm: How to balance the shortest path with dynamic factors such as safety and congestion levels?

Solution Process

Step 1: Design of Information Transmission Mechanism

Problem: Choosing between centralized or distributed information transmission methods.
Principles:

  • Centralized (e.g., monitoring center + broadcast): Information is unified and authoritative but carries a high risk of single-point failure and requires infrastructure support.
  • Distributed (e.g., verbal communication between individuals, mobile phone communication): High redundancy, but information may be distorted or have uneven delays.
    Trade-offs:
  • Hybrid model: Prioritize the central system for critical global information (e.g., exit status), with local information supplemented distributively (e.g., congestion ahead).
  • Example: In a fire, broadcasts notify of the main exit closure, while individuals perceive alternative exit routes by observing neighbors' behavior.

Step 2: Optimization of Information Update Frequency and Delay

Problem: Excessive update frequency strains communication resources; insufficient frequency leads to outdated information.
Analysis Methods:

  1. Define information validity period:
    • Static environment (e.g., building structure): Long information validity, low-frequency updates (e.g., every few minutes).
    • Dynamic environment (e.g., smoke diffusion): Short information validity, requiring high-frequency updates (e.g., every second).
  2. Delay impact modeling:
    • Use queuing theory to calculate information processing delay, ensuring the update interval is less than the information expiration time.
    • Example: If smoke spreads at 1 m/s and individual movement speed is 1.5 m/s, the update interval should be less than the time threshold for an individual to reach the hazardous area.

Step 3: Setting Re-planning Trigger Conditions

Problem: When to trigger path re-planning? Blind re-planning may cause oscillations (e.g., frequent route switching).
Strategies:

  1. Event-driven:
    • Global events: Exit closure, expansion of hazardous area → immediate forced re-planning.
    • Local events: Individual detects that ahead density exceeds a threshold (e.g., 4 persons/m²) or speed drops below a critical value (e.g., 0.2 m/s).
  2. Periodic checks:
    • Evaluate current path efficiency at fixed intervals (e.g., every 5 seconds). If the expected time increases by more than 20%, trigger re-planning.
  3. Predictive triggering:
    • Use machine learning to predict congestion trends and re-plan in advance (e.g., if the probability of path congestion within the next 10 seconds >80%).

Step 4: Design of Path Re-planning Algorithm

Problem: How to generate new paths that balance distance, safety, and real-time congestion?
Algorithm Selection:

  1. Improved traditional graph search algorithms:
    • A* Algorithm: Set the cost of dynamic obstacles to infinity and update the heuristic function in real-time (e.g., considering smoke concentration weight).
    • D* Lite Algorithm: Suitable for scenarios with partially unknown environments, incrementally updates paths to reduce computational load.
  2. Multi-objective optimization:
    • Objective functions: Minimize time, maximize safe distance, minimize congestion level.
    • Solutions: Pareto frontier analysis or weighted sum method (e.g., total cost = distance × 0.5 + congestion level × 0.3 + hazard coefficient × 0.2).
  3. Reinforcement learning-based:
    • Train agents to learn dynamic obstacle avoidance strategies in simulated environments. States include location, neighbor density, and distance to hazards; actions involve direction choices.

Step 5: System Integration and Simulation Verification

Problem: How to verify the effectiveness of the mechanism?
Methods:

  1. Simulation tools: Use platforms such as AnyLogic, FDS+Evac, and inject dynamic events (e.g., random exit closures).
  2. Evaluation metrics:
    • Total evacuation time, average individual delay, number of path switches (stability), and hazard exposure time.
  3. Parameter tuning:
    • Adjust update frequency and trigger thresholds, observe changes in metrics, and find the optimal balance point.

Summary

Dynamic path re-planning requires integrating information update mechanisms with intelligent algorithms. Key points include:

  • Information reliability: Reduce single-point failure risk through hybrid communication models.
  • Response timeliness: Adaptively adjust update frequency based on environmental dynamics.
  • Algorithm adaptability: Combine multi-objective optimization and real-time prediction to avoid local optima or oscillations.
    In practical applications, parameters need to be repeatedly calibrated through simulations to ensure system robustness in uncertain environments.