An Interface for Programming Verifiable Cognitive Agents in ROS
The Agent-ROS interface is designed for integrating cognitive agents into robotic systems developed using ROS. Programming cognitive agents to autonomously control robots allows us to separate high-level control from low-level operations. This improves modularity, giving us a layer of explainability, and makes it possible to formally verify the core of the robot that is responsible for making high-level decisions.
Robotic applications in industry are generally hard to formally verify due to the amount of low-level code. By adding cognitive agents, through the use of our interface, we can improve usability, modularity, and reliability of the application.
Opportunities include using the tool publicly available and providing complex use cases that can further push and improve our approach.
Source code is available here.