Prototyping and Simulation of Robot Group Intelligence using Kohonen Networks.

Area of Science:

• Robotics
• Artificial Intelligence
• Network Engineering

Background:

• Intelligent agents, like robots, can form ad hoc networks for dangerous scenarios, such as disaster relief.
• Current robot design and manufacturing processes can be costly and inefficient.
• There is a need for effective simulation platforms to test robot capabilities before physical deployment.

Purpose of the Study:

• To prototype and build a computer simulator for robot kinetics, unsupervised learning, and group intelligence in ad hoc networks.
• To model individual robots with attributes and methods defining their states and actions.
• To test the learning capabilities and collaborative behavior of robot groups in complex scenarios.

Main Methods:

• Object-oriented modeling of individual robots with defined attributes and methods.
• Implementation of unsupervised learning using Kohonen networks for group intelligence.
• Simulation of robot ad hoc networks in scenarios of varying complexity.
• Testing of robot group collaborative behavior on complex terrains.

Main Results:

• The simulator successfully modeled robot kinetics, unsupervised learning, and group intelligence.
• Simulations demonstrated that simple, reliable, and affordable robots can form effective ad hoc networks.
• A group of simulated robots exhibited highly collaborative behavior on complex terrain.
• The simulation platform proved effective for testing robot capabilities.

Conclusions:

• The developed simulator provides a valuable platform for testing individual and group robot capabilities.
• This approach has the potential to reduce costs and improve the efficiency of robot design and manufacturing.
• The findings support the use of intelligent agents in dangerous environments and highlight the potential of swarm robotics.