This paper presents a multi-sensor positioning and navigation system that leverages cost-effective commercial-grade sensors for GNSS-challenging urban and indoor environments. The system fuses onboard motion sensor data with LiDAR point clouds registered to high-accuracy 3D digital maps to achieve sustained decimeter-level positioning. Key contributions include accurate LiDAR scans geo-referencing with motion compensation, efficient map-to-map registration, and an effective decentralized fusion. Real-world driving data from downtown Kingston, Ontario, Canada, and a high-accuracy 3D city geodatabase were used to examine the proposed methods’ performance and benefits. Results demonstrate the efficacy of the proposed technique, achieving accurate positioning with an average RMSE of 20cm horizontally and 13cm vertically, and a sustainable positioning sub-meter level of positioning accuracy 100% of the time. The proposed method was also able to sustain high precision positioning in such GNSS-denied environments with position errors of less than 50cm for 96.8% of the time and less than 30cm for 91% of the time. The performance achieved demonstrates that the proposed system is a practical option for the positioning and navigation of self-driving cars and has the potential for cooperative mapping and updating 3D city maps.