Roadside perception systems, also known as roadside units (RSUs), are critical in Vehicle-to-Everything (V2X) applications, yet spatio-temporal asynchrony between multiple sensors severely compromises the accuracy of fusion. In this paper, a spatio-temporal synchronization method for millimeter-wave (MMW) radar and camera fusion is proposed, integrating target matching based on dynamic time warping (DTW) with spatio-temporal parameter estimation. Leveraging the advantages of DTW in time-series alignment to calculate the similarity between radar and visual trajectories enables target matching and parameter estimation in sparse scenes. This method was validated on a real-world dataset containing over 30 pedestrian trajectories, covering scenarios with varying densities ranging from one to six pedestrians. The results indicate a temporal offset of 0.116 s between the camera and radar. Following synchronization, the average spatial deviation decreased from 1.4358 to 0.1074 m in the x-direction (i.e., across the road) and from 3.0732 to 0.1775 m in the y-direction (i.e., along the road). Consequently, this method provides an efficient solution for deploying roadside perception systems in sparse traffic environments.

Light Detection and Ranging (LiDAR) three-dimensional (3D) object detection degrades under point sparsity, outliers, coordinate noise, and calibration drift, yet detector evaluation remains largely limited to clean benchmarks. This study focuses on sensing robustness rather than detector redesign. We introduce Bounded Graph Conditioning (BGC)—a deterministic pre-voxelization front-end that applies k-nearest-neighbor (kNN) neighborhood averaging with bounded residual correction upstream of an unchanged detector backbone. BGC is evaluated together with a reproducible sensor-degradation stress protocol and a risk-constrained operating-boundary analysis. Experiments on KITTI with PointPillars, SECOND, and Voxel R-CNN show that BGC most clearly improves retained detection quality and feasible operating coverage under strong noise and strong outlier stress; gains under other degradation types are smaller and backbone-dependent. In the primary score-level box-disjoint calibration/test evaluation on SECOND, maximum feasible coverage at a target risk bound of $0.2$ improves from $0.0754$ to $0.1374$ under strong noise ($\sigma =0.10$ m) and from $0.1323$ to $0.1591$ under strong outliers ($p=0.10$); a cross-backbone check on Voxel R-CNN confirms the same direction ($0.1860\to 0.2864$). Comparison with traditional filtering (SOR and ROR) reveals complementary strengths across fault types. A range-adaptive BGC variant that adjusts parameters per distance bin further improves performance under mixed unknown faults, spherical-coordinate noise, and on a dataset-matched nuScenes validation (adaptive BGC mAP/NDS: $0.2687/0.4493$ vs. baseline $0.2471/0.3846$ under strong noise). Severe translation drift collapses all configurations to full rejection, exposing an explicit sensing boundary beyond the reach of local conditioning. These results support BGC as a practical sensor-side robustness enhancement under the studied degradation protocol, with conditional rather than universal applicability across backbones and fault types.

Original high-definition radar data contains rich environmental information, including distance, Doppler velocity, and azimuth. However, extracting robust features from such sparse and noisy frequency-domain data remains a challenge. To address this issue, this paper proposes an improved multi-task network, the Efficient Feature Aggregation with Balanced Attention Radar Network (EFA-RadNet). This network introduces the VoVNetV2 architecture into the field of raw radar perception and effectively preserves feature diversity across different receptive fields through a One-Shot Aggregation (OSA) module, avoiding signal aliasing. In addition, we propose an attention mechanism module, Balanced effective Squeeze–Excitation (B-eSE), which is better suited for sparse radar processing and effectively addresses the problem of weak target loss in the radar spectrum. Experiments on the RADIal dataset show that our EFA-RadNet achieves excellent target detection performance while also attaining optimal accuracy in free space segmentation.