Sensor Fusion for Autonomous Vehicle Perception Under Adverse Weather Conditions

Autonomous vehicles employ a suite of complementary sensors—such as radar, LiDAR, cameras and GNSS—to perceive their surroundings and make driving decisions. Adverse weather conditions, including rain, fog, snow and low‐light scenarios, introduce attenuation, scattering and noise that can degrade the performance of individual sensors. Sensor fusion seeks to mitigate these vulnerabilities by integrating data from multiple modalities, thereby enhancing reliability and robustness. Modern fusion frameworks range from classical probabilistic filters that align and fuse radar range measurements with LiDAR point clouds to deep‐learning architectures that learn spatio‐temporal correlations across camera images and depth returns. Key challenges include time synchronisation, calibration drift under temperature changes, dynamic weighting of sensor outputs when one modality is compromised, and real‐time processing constraints. Progress in computational hardware and efficient algorithms now permits near‐sensor fusion, where preliminary data association and noise filtering occur at the edge, reducing latency. Globally, improved fusion strategies are critical for safe operation in regions prone to heavy precipitation or snow, enabling applications from urban mobility in temperate climates to long-distance transport in polar environments.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Sensor Fusion for Autonomous Vehicle Perception Under Adverse Weather Conditions publication trend

The graph below shows the total number of articles in sensor fusion for autonomous vehicle perception under adverse weather conditions across all publications each year (not limited to Nature Index journals).

Technical terms

Sensor fusion: The process of combining data from multiple sensors to produce a more accurate, reliable estimate of the environment than would be possible from any individual sensor.
LiDAR: Light Detection and Ranging; an optical sensing method that measures distance by timing the return of laser pulses to generate a three-dimensional point cloud.
RADAR: Radio Detection and Ranging; a sensing technique that emits radio waves and analyses their echoes to determine the range, velocity and angle of objects.
Point cloud: A set of data points in three-dimensional space representing the external surfaces of objects, typically produced by LiDAR sensors.
GNSS: Global Navigation Satellite System; a constellation of satellites delivering geospatial positioning with time synchronisation capabilities critical for sensor alignment.

References

1. An Overview of Autonomous Vehicles Sensors and Their Vulnerability to Weather Conditions. Sensors (2021).
2. Predicting the Influence of Rain on LIDAR in ADAS. Electronics (2019).
3. A Scalable and Accurate De-Snowing Algorithm for LiDAR Point Clouds in Winter. Remote Sensing (2022).