Simulator and Transplantable Control Architecture Development for an Electric Autonomous Vehicle

Developing autonomous vehicles requires a strong emphasis on safety-critical tasks. An effective approach for this challenge is to create a simulator with precise vehicle parameters. This allows control algorithms to be safely developed and tested in a virtual environment, leading to a shorter development period. This paper proposes the procedure for developing a simulator for an autonomous vehicle in ROS Gazebo. System identification based on Dynamic Mode Decomposition with Control (DMDc) is conducted for vehicle parameter characterization. A controller framework is proposed with swappable control algorithms and control targets among simulators and real vehicles. The simulator’s and real vehicle’s behavior triggered by the same control input is compared for simulator effectiveness and accuracy validation. Different control algorithms are applied to vehicle GNSS-INS based navigation development. The demo video can be found at: https://youtu.be/S_OjyMRzqWk