An Application of Self-Organizing Map for Multirobot Multigoal Path Planning with Minmax Objective

Area of Science:

• Robotics
• Artificial Intelligence
• Computational Intelligence

Background:

• Multigoal path planning is crucial for mobile robots operating in complex environments.
• The Multiple Traveling Salesman Problem (MTSP) with a minmax objective is a relevant model for coordinating multiple robots to visit multiple locations.
• Integrating unsupervised learning with path planning presents challenges in handling obstacles and ensuring collision-free trajectories.

Purpose of the Study:

• To adapt Self-Organizing Maps (SOM) for solving the robotic Multiple Traveling Salesman Problem (MTSP) with a minmax objective.
• To address the challenge of determining collision-free paths required during the unsupervised learning phases of SOM.
• To verify the suitability of the proposed SOM approach for cooperative inspection tasks using mobile robots.

Main Methods:

• Application of Self-Organizing Map (SOM) to the Multiple Traveling Salesman Problem (MTSP) in a polygonal domain.
• Utilization of approximate shortest path algorithms to determine collision-free paths for neuron-city distance evaluation and adaptation.
• Formulation of a cooperative inspection task as an MTSP-Minmax problem.

Main Results:

• The proposed SOM approach effectively solves the robotic MTSP by incorporating approximate shortest path calculations.
• The method successfully determines collision-free paths necessary for both winner selection and neuron adaptation phases in SOM.
• The SOM-based solution achieved competitive results when compared against the GENIUS combinatorial heuristic for the cooperative inspection task.

Conclusions:

• The integration of approximate shortest path methods with SOM provides a viable solution for robotic multigoal path planning.
• The Self-Organizing Map is a suitable tool for coordinating groups of mobile robots in complex planning scenarios.
• This research opens new avenues for applying SOM in robotic planning and multi-agent coordination problems.