This project focuses on developing a sim-to-real autonomous navigation stack for drive-by-wire vehicles using ROS2, integrating AMCL-based localization, NAV2 path planning, and a custom Stanley lateral controller with proportional longitudinal control to achieve closed-loop autonomous driving in mapped environments.
Sim-to-Real Autonomous Navigation Stack for Drive-By-Wire Vehicles
What is it?
Why is it needed?
Bridging the gap between simulation and real-world autonomous vehicle deployment requires validated software stacks tested on production vehicle platforms. A fully integrated ROS2 autonomy stack on a drive-by-wire vehicle provides a realistic testbed for evaluating localization accuracy, controller performance, and sensor fusion under real dynamic conditions.
What is CAST lab doing to improve it?
CAST Lab is developing and validating a complete autonomous navigation pipeline on the Lincoln MKZ, including custom NAV2 controller plugins, Stanley path tracking, and a sensor fusion architecture combining lidar-based AMCL localization with wheel odometry. Current research focuses on improving localization robustness, controller stability across varying speeds, and seamless transition from simulation to real vehicle deployment.