Economic Platform for Action Model Sharing Based on Blockchain in Mobile Robot Networks

Abstract

Data sharing among robots will be an important issue in a future robotic society including many kinds of networked robots. This study proposes a model-sharing system for autonomous mobile robot networks. The model in this study means an action model trained by deep reinforcement learning for mobile robot navigation. The proposed system assumes that the mobile robots in the network can share useful and efficient models among robots. The proposed system uses the Ethereum blockchain as a platform to consider the value and ownership of action models. The owners of the models can receive payments from robots that use the shared models for autonomous navigation. The proposed system finally aims to build a future robotic economic system that includes data generation and accumulation, data sharing, and currency transactions among model owners and user robots. This paper introduces the autonomous navigation method of switching action models according to the navigation environment, the data-sharing system using the Ethereum blockchain, and the flow of autonomous navigation using the proposed system. Simulation results shows that the proposed system can achieve navigation of multiple mobile robots through model and currency transactions among Ethereum network participants.