HyperNEAT Capture the Flag

Multi agent learning of capture the flag using HyperNEAT(Hyper cube based Neuroevolution of augmenting topologies) with heterogeneous agents.

As shown the red team learned to defend, but does not learn to capture. The blue team appears to have learned about both flags, but is only able to capture the green flag. The green team learns to separate the red and the blue teams. However, each team seems to be working as a team rather than independently.

My intent is to learn if competing teams will eventually learn when it is in there best interest to work together against a competing team. Currently the ai is not that smart.

Check out my website drew.freehostingcloud.com

Category:

Science & Technology

Tags:

• neural
• network
• genetic
• algorithm
• HyperNEAT
• NEAT
• ai
• Artificial
• Intelligence
• machine
• learning
• multi
• agent