Tests on a standard autonomous-driving dataset found that using both types of sensors worked better than relying on either one alone.
Researchers developed a system that combines camera images with radar data to detect objects in three dimensions around an autonomous vehicle. It converts visual information into a 3D form before combining it with radar-based features.
In tests on the nuScenes dataset, the combined system performed better than systems using only a camera or only radar. Its performance was also competitive with other leading camera–radar fusion methods, although the abstract does not report numerical results.
The researchers used a deep neural network to extract features from camera images, then transformed those features into 3D space with a Cross-Domain Spatial Matching step. They fused the transformed camera information with radar-derived features to create a shared 3D representation of objects.
On the nuScenes dataset, the method outperformed single-sensor camera and radar systems. It achieved competitive performance compared with other leading sensor-fusion methods. The abstract does not give specific accuracy or detection scores.
Cameras provide detailed visual information, while radar supplies complementary measurements that can help represent objects in three dimensions. The results suggest that combining the two data sources can improve object detection compared with using either sensor alone, an important part of an autonomous vehicle’s perception system.
The evidence comes from experiments on the nuScenes dataset, comparing the camera-and-radar system with single-sensor baselines and other sensor-fusion methods. The abstract does not provide numerical results, and it does not establish how the approach performs in other datasets, in real-world driving, or in a deployed vehicle.
Scientific Reports · 2026 · DOI: 10.1038/s41598-026-66505-1
Authors: Daniel Dworak, Mateusz Komorkiewicz, Paweł Skruch, Jerzy Baranowski
Institutions: AGH University of Krakow
Researchers developed an aerial robot that uses touch to refine its position and perch on structures it has not encountered before. The system succeeded in more than 99% of simulated trials and completed 26 real-world perching trials despite corrupted position estimates.
A modeling study of Belgium’s 2045 energy system found that all near-optimal low-carbon pathways met the climate target but exceeded limits for ecotoxicity and particulate matter under the most permissive assessment approach. The pathways had similar costs but differed in whether they relied mainly on electrification, hydrogen, renewable power capacity or a mix of strategies.
Researchers developed a way to calculate traffic sensitivities efficiently in road networks with different vehicle classes, merges, diverges and congestion spillback. Tests on a sample network found the approach was particularly efficient as the number of control parameters increased.
