Evolution of a Communication Code in Cooperative Tasks (2012)
Author: Aditya Rawal, Padmini Rajagopalan, Risto Miikkulainen, Kay Holekamp
Communication through vocalizations is used by spotted hyenas and chimpanzees for coordination during hunting and for raising alarm calls in defense (Bullinger et al., 2011; Holekamp KE and BL, 2007). Vocal signals are omni- directional and are therefore more effective than visual communication in these situations. In cooperative tasks, agents use these signals to pro-actively exchange information for common good. A simulated predator-prey domain is considered in this paper - where multiple predator agents exchange real valued messages as an approximation of vocalization in nature. In artificial intelligence, the problem of coordination among multiple predator agents during prey capture is hard because of the non-Markovian environment (Panait and Luke, 2005). Experiments are carried out in this paper to show how information exchange through messaging can make the environment less non-Markovian and improve predator team performance during cooperative hunt. The values of these messages are analyzed to study the emergence of a common language among the predator agents. The results in this paper also provide an insight on the constraints under which language evolves in nature.

Video description

2 predators (green and blue blocks) with messaging/signaling pursuing a fast moving prey (pink block) in a toroidal world . Prey has the same speed as the predators (as against 0.75x in the paper) and it follows a fixed policy of moving away from the closest predator. Therefore no single predator can catch the prey individually by greedily chasing it. Predators can sense the prey position but do not sense each other's location directly. However, the predators evolve to send real valued messages to each other which enables cooperation between them. Note, prey is caught multiple times during the simulation (at times 0:21, 0:37, 0:55, 1:19, 1:36, 2:12). The prey reappears at a random position in the world after being caught. The best team after 1500 generations is selected for demonstration here. The predators have evolved a strategy of attacking the prey from opposite directions and catch it.

Kay E. Holekamp Collaborator holekamp [at] msu edu
Evolution of a Communication Code in Cooperative Tasks 2012
Aditya Rawal, Padmini Rajagopalan, Risto Miikkulainen and Kay Holekamp, In Artificial Life (13th International Conference on the Synthesis and Simulation of Living Systems), East Lansing, Michigan, USA 2012.