Swarm Contracts: Smart Contracts in Robotic Swarms with Varying Agent Behavior

Abstract

Multi-agent robotic systems are becoming pervasive in many real-world applications from search and rescue missions to future household robotic appliances that might need to work together to achieve specific tasks. We propose and implement a collaborative environment for secure communication of robotic agents in a prototype agent system that mimics the interactions between agents of varying behaviors using special-purpose smart contracts titled "Swarm Contracts." This paper describes how Swarm Contracts and blockchain technologies increase the interaction efficacy between agents by providing a more trusted information exchange to reach consensus under trustless conditions, assess agent productivity, allocate plans and tasks to deploy distributed solutions, and carry out joint missions. All these features are encapsulated in Swarm Contracts, making the decentralized applications that use them a viable alternative to centralized command and control applications that are pervasive in multi-agent robotics applications of today. We have evaluated the utility of the developed Swarm Contracts in adversarial settings and report the results that are very promising for future applications of such decentralized heterogeneous robotic agent interactions.