Architecture of autonomous vehicle simulation and control framework

Abstract

With the current advances in Autonomous Navigation Systems for Self Driving Vehicles, there is a need for specialized and focused software architecture for ease in simulating the vehicle behavior and associated modules such as perception, drive by wire system, obstacle detection and path planning. In addition, testing algorithms on real vehicles during development takes enormous effort, time and also contains risk to the vehicle and its corresponding surroundings. Having a software architecture that is very similar to the target autonomous system combined with a fairly accurate simulator, reduces this risk and eases the development process. In this paper, we propose an integrated software architecture using ROS and Gazebo that is compliant to JAUS standards, which is scalable and easy to use. The proposed architecture uses Gazebo that is equipped with the Open Dynamics Engine providing realism to the simulation along with sensors perception, path planning and vehicle control using a Drive-by-Wire system. The results of using the proposed software architecture are shown for Autonomous and Tele-operated navigation missions on an all-terrain vehicle.