GNSS-Denied Simultaneous Localization and Mapping (SLAM) with 4D Imaging Radar

Based on previous project “Autonomous Navigation using mmWave Radar Sensors“, we further enhanced the system by upgrading it to support up to four 4D imaging radars. The setup mainly uses three Altos radars, while also being compatible with the other Texas Instruments AWR2944 FMCW radar sensor. Even if one of the sensors fails, the system can still reliably estimate the ego motion, ensuring robust and continuous operation.

GNSS-denied environments remain one of the major challenges in autonomous navigation, where traditional GNSS-based positioning becomes unreliable or completely unavailable. In this work, we demonstrate how SLAM (Simultaneous Localization and Mapping) can be achieved using 4D imaging radar — an advanced sensing technology capable of providing range, velocity, azimuth, and elevation information in real time.

Unlike conventional LiDAR or camera-based systems, radar-based SLAM offers strong robustness in harsh environments such as fog, rain, dust, or low-light conditions. The following video showcases how modern 4D imaging radar systems can estimate motion, build environmental maps, and maintain accurate localization even in GNSS-denied scenarios.

We did the outdoor experiment on 8th January 2026 in Kirchberg, Luxembourg. It was a rainy day, as usual in Luxembourg, which made the conditions more challenging and further highlighted the robustness of radar-based perception for all-weather autonomous navigation. In addition, we drove through tunnels to analyze the condition of GNSS signal loss and instability.

This following video presents an analysis of a scenario where Radar SLAM continues to operate effectively in environments where GNSS signals are lost or unreliable, such as inside tunnels. It highlights how radar-based localization can maintain robustness in challenging conditions where satellite navigation systems fail or become unstable.

Mobile Robot Setup for University On-campus SLAM

For on-campus Radar SLAM for further analysis, the above mobile robot setup has been designed and upgraded instead of using TurtleBot 4 Standard robot, in order to better support radar sensors integration and more outdoor SLAM testing.