Multiagent deep reinforcement learning based self-organizing system design

Graduate Researcher: 
Hao Ji

Faculty Adviser:   
Prof. Yan Jin

Self-organizing systems (SOS) are able to perform complex tasks in unforeseen situations with adaptability. Previous work has introduced field-based approaches and rule-based social structuring for individual agents to not only comprehend the task situations but also take advantage of the social rule-based agent relations in order to accomplish their overall tasks without a centralized controller. Although the task fields and social rules can be predefined for relatively simple task situations, when the task complexity increases and task environment changes, having a priori knowledge about these fields and the rules may not be feasible. In this paper, we propose a multi-agent reinforcement learning based model as a design approach to solving the rule generation problem with complex SOS tasks. A deep multi-agent reinforcement learning algorithm was devised as a mechanism to train SOS agents for acquisition of the task field and social rule knowledge. Learning stability, team differentiation and robustness property of this learning approach were investigated with respect to the changing team sizes and box dimensions. Through a set of simulation studies on a box-pushing and self-assembly problem, the results have shown that there is an optimal range of number of agents that achieves good learning stability; agent teams learn to differentiate with changing team sizes and box dimensions; robustness of knowledge and robustness to situation change vary with changing box dimensions..

Related Publications:

Ji, H. and Jin, Y. “Design Self-Organizing Systems with Deep Multiagent Reinforcement Learning”, ASME 2019 International Design Engineering Technical Conferences & Computers and Information in Engineering Conference, IDETC2019-98268, Aug. 18-21, 2019, Anaheim, CA, USA.

Ji, H. and Jin, Y. “Modeling Trust in Self-Organizing Systems with Heterogeneity”, ASME 2018 International Design Engineering Technical Conferences & Computers and Information in Engineering Conference, IDETC2018-86006Aug. 26-29, 2018, Quebec City, Quebec, Canada.

Ji, Hao, Jin, Yan. “Adoption of Social Rules in Teams of Different Sizes” in Engineering Management Reviews, 2017, 6(1), 6-15. doi: 10.14355/emr.2017.0601.002

Read more about this project.