Radar-Based Localization and Mapping Techniques

Radar-based localization and mapping rely on radio frequency signals to determine the position and geometry of a platform within an environment. These techniques leverage the robustness of radar waves against adverse weather and lighting conditions, offering a complementary or alternative solution to optical sensors and LiDAR. Key methods include Simultaneous Localization and Mapping (SLAM), in which feature extraction, motion estimation and loop closure jointly build a consistent map and trajectory; radar odometry, which incrementally estimates displacement through Doppler and range measurements; and hybrid systems that fuse radar data with GNSS or inertial sensors to impose global constraints. Advances in signal processing, sensor fusion and machine learning have improved feature stability, noise rejection and cross-domain generalisability. Modern systems employ continuous‐wave (CW) or frequency‐modulated continuous‐wave (FMCW) radars, capable of measuring range, velocity and sometimes azimuthal angle of targets. Applications span autonomous vehicles, drones and robotics for reliable navigation in urban environments, underground tunnels and all-weather scenarios. The global significance of radar mapping is underscored by its utility in safety‐critical operations—from traffic management to disaster response—where optical sensors may fail. Concrete examples include urban SLAM using spinning radars, underground localisation with ground‐penetrating radar and tunnel exploration aided by false-detection mitigation. A common trend is the emergence of tightly coupled sensor fusion architectures and data‐driven algorithms that enhance accuracy, robustness and interpretability.

Research from Nature Portfolio

Recent studies have proposed a deep learning framework for robust all-weather positioning by fusing visual and radar data within a self-supervised architecture. This geometry-aware model applies attention mechanisms to weigh sensor inputs, producing reliability masks that filter out unreliable measurements under rain, fog or snow. The approach demonstrates strong cross-domain generalisability, adapting from day to night conditions without additional labelling. A game-theoretic analysis of model predictions reveals distinct failure modes for each modality, offering insights into system interpretability and resilience. Such work brings radar one step closer to a dependable localisation solution for autonomous driving in challenging environmental scenarios.

Radar-Based Localization and Mapping Techniques publication trend

The graph below shows the total number of articles in radar-based localization and mapping techniques across all publications each year (not limited to Nature Index journals).

Technical terms

Simultaneous Localization and Mapping (SLAM): A process that concurrently builds a map of an unknown environment and estimates the position of the sensor within it.

Frequency Modulated Continuous Wave (FMCW) radar: A radar type that emits a continuously varying frequency signal to measure range and velocity of targets with high resolution.

Radar Cross Section (RCS): A measure of an object’s detectability by radar, representing the effective reflective area presented to the radar beam.

Global Navigation Satellite System (GNSS): A satellite-based system that provides geospatial positioning with global coverage and time synchronisation.

Loop Closure: A method in SLAM that recognises previously visited locations to reduce accumulated drift and ensure map consistency.

References

1. HGCM: A hybrid radar feature and GNSS based continuous mapping method. Expert Systems with Applications (2025).
2. Deep learning-based robust positioning for all-weather autonomous driving. Nature Machine Intelligence (2022).
3. Localization and Mapping Using Only a Rotating FMCW Radar Sensor. Sensors (2013).
4. Autonomous Navigation in Inclement Weather Based on a Localizing Ground Penetrating Radar. IEEE Robotics and Automation Letters (2020).
5. False Detections Revising Algorithm for Millimeter Wave Radar SLAM in Tunnel. Remote Sensing (2023).