Advancing perception algorithms for autonomous vehicles across diverse domains, including urban roads, off-road terrains, and high-speed racing by enhancing LiDAR-based vehicle detection, multi-modal sensor fusion, and robust GNSS-denied localization methods. Tested on platforms like autonomous racecars, off-road defense vehicles, and motorcycles, this work has a proven track record of enhancing safety, adaptability, and real-time performance of existing perception systems.