Leaderboards

To evaluate the performance of each of the agents discussed in agents, we propose a evaluation environment which is a 2v2 team environment with 36 wooden blocks and having no rigid blocks or powerups.

Every one of these agents are made to compete in the following contests for 100 episodes each and the win, loss and the tie percentage are plotted.

  1. Agent and static agent vs Two random agents

  2. Agent and static agent vs Random agent and static agent

  3. Agent and static agent vs Two static agents

  4. Agent and static agent vs Simple agent and Static agent

  5. Two Agents vs two Simple agents