Refining Roadside 4D Radar Vehicle Classification Using Point-Cloud Dispersion
Authors: Mohammed Alsakabi, Andreia Figueiredo, António Almeida, John Dolan, Miguel Luís, Pedro Rito, Ozan Tonguz, Susana Sargento
Organizations: Carnegie Mellon University, Pittsburgh, USA · Departamento de Electrónica, Telecomunicações e Informática, Universidade de Aveiro, Portugal · Instituto de Telecomunicações, Universidade de Aveiro, Campus Universitário de Santiago, 3810-193, Aveiro, Portugal · Instituto de Telecomunicações, Instituto Superior Técnico, Universidade de Lisboa, 1049-001 Lisboa, Portugal
Roadside sensing infrastructure is essential for Intelligent Transportation Systems (ITS), but cameras and legacy sensors are limited in range, weather robustness, and depth accuracy. 4D radar complements these modalities by resolving range, azimuth, elevation, and radial velocity under adverse conditions, though its sparse, noisy point clouds make reliable classification difficult. We present a roadside multi-sensor platform combining a 4D radar and stereo depth camera under synchronized acquisition, used to collect a real-world dataset. Our analysis reveals a substantial class-distribution discrepancy between modalities, most notably a strong radar confusion between passenger cars and motorcycles (49.01% and\ 48.60% of the dataset). To address this, we propose a Neyman-Pearson-based binary classifier on the temporally-smoothed 2D positional standard deviation of radar point cloud clusters, reducing classification error from 52.1% to 42.7%. This lightweight and edge-deployable approach improves radar-only classification reliability and lays the foundation for future camera-radar fusion.
Figures & tables
Fig. 1: Architecture of the sensing platform.
Field
Description
Unit
x
Cartesian X coordinate
m
y
Cartesian Y coordinate
m
z
Cartesian Z coordinate
m
range
Measured target distance
m
azimuth
Horizontal angle
rad
elevation
Vertical angle
rad
TABLE I: Point cloud attributes stored in each .pcd file.
Field
Description
Unit
timestamp
Object observation timestamp (UNIX epoch)
s
objectID
Unique radar tracking identifier
–
classification
Object class label (1=person, 2=bicycle, 4=motorcycle, 5=car, 6=bus, 7=truck)
–
confidence
Classification confidence
%
x
Local X position in radar frame
m
y
Local Y position in radar frame
m
TABLE II: Attributes stored in the radar SQLite Object List database.
Field
Description
Unit
timestamp
Object observation timestamp (UNIX epoch)
s
objectID
Unique camera tracking identifier
–
classification
Object class label (1=person, 2=bicycle, 4=motorcycle, 5=car, 6=bus, 7=truck)
–
confidence
Classification confidence
%
latitude
Global latitude of object
deg
longitude
Global longitude of object
deg
TABLE III: Attributes stored in the camera SQLite Object List database.
Fig. 2: Roadside deployment of the radar and stereo camera.
Fig. 3: Bird’s-eye view of radar and camera object observations
Object Class
Total object observations
Percentage (%)
Passenger Car
12,781
49.01
Motorcycle
12,675
48.60
Bicycle
562
2.15
Truck
63
0.24
Total
26,081
100.00
TABLE IV: Distribution of Classified Radar object observations
Object Class
Total object observations
Percentage (%)
Passenger Car
2,971
85.69
Truck
185
5.34
Person
184
5.31
Bicycle
77
2.22
Motorcycle
40
1.15
Total
3,467
100.00
TABLE V: Distribution of Classified Camera object observations
Fig. 4: Comparison of percentages between Radar and Camera object observations
Fig. 5: Radar object observations across different ranges
Fig. 6: Distribution of 2D positional standard deviation σ for passenger cars and motorcycles. The dashed red line indicates the Neyman-Pearson decision threshold η=0.14 m.
Fig. 7: Class distributions of the 2D positional standard deviation σ after temporal averaging with window W=100 . The dashed line indicates the optimal threshold η∗=0.16 m.
Fig. 8: Total classification error rate as a function of the decision threshold η . The optimal threshold η∗=0.16 m minimizes the error rate at 42.7% .
4D automotive radar is indispensable for autonomous driving due to its low cost and robustness, yet its point cloud sparsity challenges 3D object detection. Existing 4D radar-camera fusion methods focus on complex fusion strategies, trading inference speed for marginal gains. This trade-off hinders real-time deployment due to heavy computation on dense feature maps. In contrast, feature extraction from sparse radar points is less time-consuming but remains under-explored. This work uncovers that simply enhancing radar feature extraction can achieve comparable or even higher performance than elaborate fusion modules, while maintaining real-time performance. Based on this finding, we propose RCGDet3D, which centers on radar feature encoding and simplifies multi-modal fusion. Its encoder inherits from the efficient Gaussian Splatting-based Point Gaussian Encoder (PGE) in RadarGaussianDet3D with two key improvements. First, the Ray-centric PGE (R-PGE) predicts Gaussian attributes in ray-aligned coordinate systems before unifying them to Bird's-Eye View (BEV) space, significantly improving geometric consistency and reducing learning difficulty by decoupling the coordinate transformation from representation learning. Second, a Semantic Injection (SI) module incorporates visual cues from images, producing more geometrically accurate and semantically enriched radar features. Experiments on View-of-Delft (VoD) and TJ4DRadSet show that RCGDet3D outperforms state-of-the-art methods in both accuracy and speed, setting a new benchmark for real-time deployment.
Weiyi Xiong, Bing Zhu
School of Automation Science and Electrical Engineering, Beihang University, Beijing, P.R. China
4D millimeter-wave radar has emerged as a promising sensing modality for autonomous driving due to its robustness and affordability. However, its sparse and weak geometric cues make reliable instance activation difficult, limiting the effectiveness of existing radar-camera fusion paradigms. BEV-level fusion offers global scene understanding but suffers from weak instance focus, while perspective-level fusion captures instance details but lacks holistic context. To address these limitations, we propose SIFormer, a scene-instance aware transformer for 3D object detection using 4D radar and camera. SIFormer first suppresses background noise during view transformation through segmentation- and depth-guided localization. It then introduces a cross-view activation mechanism that injects 2D instance cues into BEV space, enabling reliable instance awareness under weak radar geometry. Finally, a transformer-based fusion module aggregates complementary image semantics and radar geometry for robust perception. As a result, with the aim of enhancing instance awareness, SIFormer bridges the gap between the two paradigms, combining their complementary strengths to address inherent sparse nature of radar and improve detection accuracy. Experiments demonstrate that SIFormer achieves state-of-the-art performance on View-of-Delft, TJ4DRadSet and NuScenes datasets. Source code is available at github.com/shawnnnkb/SIFormer.
Xiaokai Bai, Lianqing Zheng, Si-Yuan Cao +6
College of Information Science and Electronic Engineering, Zhejiang University, Hangzhou 310027, China · School of Automotive Studies, Tongji University, Shanghai 201804, China · College of Computer Science and Technology, Zhejiang University, Hangzhou 310027, China, and also with the Jinhua Institute of Zhejiang University, Jinhua 321299, China +1
4D radar has emerged as a promising next-generation sensor for improving the robustness of autonomous driving perception systems because of its stable sensing capability under adverse weather conditions. However, deploying 4D radar in embedded environments with limited hardware resources requires radar-representation preprocessing that jointly considers perception accuracy, real-time performance, and computational complexity. This paper proposes a preprocessing framework for 4D-radar-based 3D object detection. First, Percentile-based 3D Shape Preservation (P3DP) extracts point clouds from radar tensors while preserving object-shape information and suppressing noise and false alarms. Second, Multi-frame-based Noise Point Discrimination using Kernel Density Estimation (MF-KDE) improves the density and reliability of sparse radar point clouds. Finally, Embedded & NetScore (ENS) evaluates suitability for embedded deployment by jointly considering accuracy, real-time performance, adverse-weather robustness, and model complexity.
Woo-Jin Jung, Dong-Hee Paek, Jeong-Su Park +1
Cho Chun Shik Graduate School of Mobility Korea Advanced Institute of Science and Technology (KAIST) Daejeon 34051, Republic of Korea · Mechanical Engineering Research Institute Korea Advanced Institute of Science and Technology (KAIST) Daejeon 34141, Republic of Korea · Autonomous Driving Perception Technology Vanguard Team Hyundai Motor Company Seongnam-si 13529, Republic of Korea