Product Concept Development and Evaluation Using Multiagent Reinforcement Learning

Product concept development is an iterative and time-consuming task. A wide range of solutions must be developed and evaluated for the optimal result. Current methods in product concept development rely on experience of designers to explore different solutions. Reinforcement learning is a machine learning paradigm where an agent learns to make sequential decisions by interacting with the environment, receiving rewards or penalties in return for its actions. An automatic approach is introduced in this article to manage design data and knowledge in using reinforcement learning for product concept development and evaluation. A multiagent reinforcement learning method is proposed to enable different agents working and learning together in a shared design environment. The environment is formed by the design data and knowledge based on quality function deployment and axiomatic design for different agents to achieve the same objective collaboratively. The proposed method improves functionality, efficiency, and user experience of the design process in product concept development. A case study of designing a rehabilitation device verifies the effectiveness of the proposed approach.