Go to DBLP homepageBDI for Autonomous Mobile Robot Navigation
Abstract: We explored how mobile Belief-Desire-Intention (BDI) agents could navigate using path plans that are automatically generated in AgentSpeak, asking if there could be any performance advantages gained by having an agent’s path be automatically generated as a BDI plan that can be monitored, suspended and resumed in case of contingencies. To do the exploration, we used Jason BDI to design a framework to test this premise with simulated mobile robots. We further explored the navigation of mobile agents to see if such functionality should be implemented within the agent in either AgentSpeak or as an internal action, or externally in an environmental module. These agents navigated through three environments of varying complexity: a simple synchronized grid, an asynchronous grid connected via Robot Operating System (ROS), and an autonomous car simulated with AirSim connected using ROS. We demonstrated that our framework handles plan interruptions, such as preventing collisions, managing consumable resources, and updating a map when necessary while moving through an environment; that Jason BDI agents are capable of controlling autonomous mobile robots; and that the AgentSpeak language provides advantages for implementing the navigation search behaviours.
Loading