4D Radar Inertial SLAM

Throwing back to my time at ITRI, where I led a cross-functional team developing a 4D radar-inertial SLAM system on LiDAR maps.

By placing 4D imaging radar at the core and fusing it with IMU data through factor graph optimization in GTSAM, we achieved 0.26 m RMSE accuracy while showing the potential for significant cost savings compared to LiDAR-centric localization approaches.

This project demonstrated how 4D radar can serve as a robust backbone for autonomous vehicle localization — scalable, resilient, and cost-effective.

Sharing a self-driving testing clip that highlights how radar, IMU, and factor graph optimization can come together to advance the future of autonomous vehicle perception.