Keywords: multi-agent system, multi blockchains, trustworthy AI, smart contracts, polynomial computability
Abstract: A hierarchical multi-blockchain architecture for time-exact multi-agent system (MAS) ensure predictable transaction processing and verifiable smart contract execution. By leveraging he polynomial hierarchy and polynomial programming methodology, proposed framework integrates a reinforcement learning-based dual-mode data sharing mechanism tailored for embodied AI swarms that dynamically adapts communication fidelity from lightweight textual updates to high-fidelity sensory data sharing based on real-time context and resource constraints. Inspired by the Chinese social credit system, a reputation-based social credit mechanism is introduced to allow continuous assessment and reinforcement for agent reliability, enhancing trust and resilience within decentralized AI swarms. By combining dynamic stakeholder routing, temporal synchronization across global, regional, and local blockchain layers, and formal verification techniques, this approach addresses key limitations of conventional blockchains — including scalability, inter-agent coordination, and timing uncertainties — paving the way for next-generation, trustworthy MAS in complex domains such as urban management, supply chain logistics, and emergency response.
Submission Number: 9
Loading