Development of decentralized based reactive control strategy for intelligent multi-agent mobile robotics system

摘要

Multi-agent system is one of the most popular research interests in robotics nowadays. How would collections of simple robots aid us in human endeavours? Let us think about an ant colony. Ants have little capability, but when many of them are working together, they can do incredible tasks. By examining the major control architectures available in robotics area, decentralized-based reactive control architecture gives the tools required. Priority-based scheme is used in arbitration level to select which behaviour to run. It does not need to gather all information from various sensors to plan its action like deliberative control, which is also known as sensors fusion that clearly required a lot of computational time. Besides that, the mapping that the deliberate control architecture produces might not be valid at the time when changes occur between the time it gathers the information and the time it plans its actions. Instead, intelligence will arise when these simple agents work together to perform some complex task cooperatively. Our multi-agent system consists of three mobile robots that are required to transfer a large object through a restricted area. We are using a leader follower strategy to accomplish this task. Intelligence in multi-agent system in our context is that any agents have a possibility to be a leader or a follower depending on current situation of the environment. This means that all agents are homogeneously equal in hardware and software. By having this capability, it will ensure the burden of the programming task will be decreased because there will be only one program needed for the whole system. In this paper, decentralized reactive control architecture is used to develop an intelligent multi-agent cooperative strategy to perform complex task like carrying a load and navigation.