The Challenge

This project presents a flexible and dynamic simulation environment designed as a Digital Twin of a real-world multi-robot navigation system. Built using AnyLogic, the model allows real-time interaction with external clients & servers, supports dynamic configuration of maps and robot properties via JSON files, and simulates realistic behaviors such as obstacle detection, task execution, and robot coordination in environments with fixed and moving elements. The primary goal is to test and visualize robot behavior in a safe, controlled, and modifiable digital environment prior to real-world deployment.

In systems involving multiple autonomous robots navigating shared environments, collisions between robots or with static obstacles (e.g., walls, columns) can lead to significant operational disruptions. Clients require a testing platform to validate navigation and task assignment algorithms without the cost or risk of live experimentation. Furthermore, flexibility in configuration and real-time interaction with external task allocation systems is essential.

The Solution

Using AnyLogic as both a client and server, the simulation mimics the real-time behavior of a robot fleet within a configurable map. Each robot is represented as an agent and acts as both a local server (for receiving tasks) and a client (sending status updates) through Java HTTP servers and InetSocketAddress.

Main Capabilities

• Fully dynamic map and robot configuration using JSON files, allowing testing on multiple maps and scenarios without having to change anything in AnyLogic.
• Real-time communication with external APIs to receive motion plans or tasks and return robot states.
• Obstacle avoidance using Safety Zones and spatial quadrant logic.
• Visual and logical representation of nodes, links, robots, and restricted areas.
• Task execution validation through simulated collisions, obstacle detection, and battery level tracking.

Simulation Model Architecture

• Agent-Based Paradigm: The map, robots and static obstacles are modeled as agents. Additional agent classes handle task logic.
• Map Generation: The map is dynamically built in AnyLogic based on a JSON input that describes:
  • Nodes and links
  • Obstacles (circles, segments)
  • Special zones (slowdown or exclusion areas)
• Robot Profiles: A separate JSON file contains robot-specific parameters such as:
  • Dimensions, server port, platform height
  • Initial battery level and task-specific energy consumption
  • Initial position, speed & rotation speed
  • Configurable safety zones with velocity thresholds

Client-Server Interaction

• Robot as Server: Upon initialization, each robot opens a local server to receive task instructions (e.g., motion plans) from an external API and replies with success/failure status.
• Robot as Client: Each robot simultaneously acts as a client, periodically reporting its current position and battery level to the external decision-making server.

Obstacle and Collision Handling

• Quadrant System: The environment is divided into spatial quadrants to reduce the computational load of checking for nearby objects. Only adjacent quadrants are checked during movement.
• Safety Zones: Each robot has multiple configurable safety zones extending from its body. If an obstacle enters a safety zone:
  • The robot reduces speed.
  • If contact occurs with the robot body, the robot stops and flags the task as failed.
  • No Real-Time Re-Planning: Robots follow predefined motion plans from the external system. There is no dynamic rerouting; instead, failures (e.g., collisions) are reported back to the system for analysis.

Results

This simulation is not intended for performance optimization or Monte Carlo analysis. Instead, it operates in real time and is used to:

• Visually verify how robots behave when given motion plans by the external task allocation system.
• Identify potential collisions and evaluate whether the provided plans are viable.
• Serve as a testbed for the external client’s algorithms, enabling debugging and risk-free iteration.

The benefits:

• Risk Mitigation: Test robot behaviors in a safe virtual environment before deploying physical robots.
• Flexibility: Fully customizable maps, robot types, and simulation rules allow easy adaptation for future use cases.
• Real-Time Feedback: Continuous communication with external systems enables dynamic testing of task planning algorithms.
• Scalability: The architecture supports adding more robots or obstacles by updating configuration files without modifying core simulation logic.

This simulation solution offers a powerful, flexible, and interactive environment to support multi-robot systems in development. By using AnyLogic as both the engine and communication layer, and through a highly configurable agent-based design, clients can visualize, debug, and refine their autonomous robot systems efficiently and safely.