Stochastic Market Games
This addresses cooperation challenges in mixed-motive multi-agent systems, though it appears incremental as it applies market forces to known game theory problems.
The paper tackles the problem of undesirable greedy behavior in mixed-motive multi-agent systems like autonomous driving by proposing a market-based approach to incentivize cooperation, demonstrating in experiments that markets improve overall and individual agent returns.
Some of the most relevant future applications of multi-agent systems like autonomous driving or factories as a service display mixed-motive scenarios, where agents might have conflicting goals. In these settings agents are likely to learn undesirable outcomes in terms of cooperation under independent learning, such as overly greedy behavior. Motivated from real world societies, in this work we propose to utilize market forces to provide incentives for agents to become cooperative. As demonstrated in an iterated version of the Prisoner's Dilemma, the proposed market formulation can change the dynamics of the game to consistently learn cooperative policies. Further we evaluate our approach in spatially and temporally extended settings for varying numbers of agents. We empirically find that the presence of markets can improve both the overall result and agent individual returns via their trading activities.