Model-Based Autonomous Land Rover

A model-based rover platform for GPS, odometry, sensor-fusion, and autonomous waypoint-navigation tasks.

Technologies
ROSSimulinkTeensyMATLABExtended Kalman Filter
Wiring and data-flow diagram for the autonomous land rover platformSimplified runtime architechtureFinal autonomous land rover vehicle with sensors and electronics mountedpartial simulink model

Context

The rover was developed across increasingly complex autonomy tasks that validated GPS navigation, wheel-encoder odometry, range sensing, and waypoint generation under practical compute and integration constraints. The system combined a LiPo-powered electronics stack, mini-PC, microcontroller, GPS, camera, motor drivers, and manual-override path, with sensors enabled or removed depending on each task requirement to keep the platform reliable on constrained hardware.

What I contributed

I led systems integration for a five-person team and independently built roughly 80% of the rover’s GNC architecture, connecting waypoint logic, localization feedback, and speed/heading control into a closed-loop Simulink pipeline. I also configured the ROS/MAVROS communication path between the Pixhawk, onboard computer, and embedded motor controller, validated sensor and command flow, and made practical architecture tradeoffs to preserve waypoint-following reliability under onboard compute limits.