26 September 2022 in Journal of Environmental Management
11 January 2024 in Cement and Concrete Composites
16 March 2023 in IEEE Sensors Journal
This paper focuses on the challenge of pedestrian tracking using camera and millimeter wave radar in autonomous driving. Pedestrian tracking using a single sensor has inherent limitations due to the lack of comprehensive dimensionality of tracking information. Meanwhile, existing multi-sensor based tracking algorithms suffer from limited tracking accuracy by applying the fusion of projected positions. To enhance the tracking accuracy and robustness, a multi-sensor based tracking algorithm based on fused detection of millimeter-wave radar and vision is proposed, which improves the association of detection results from multiple heterogeneous sensors by utilizing newly designed back-projection and undirected graph, and finally improves the fusion detection by simultaneously utilizing a pedestrian’s appearance, local and global location information. Field tests are conducted to produce dataset, and the performance evaluation results based on the self-produced dataset have verified the superiority of the proposed algorithm over the conventional single-sensor based tracking algorithm and multi-sensor based tracking algorithms.
24 September 2021 in IEEE Transactions on Geoscience and Remote Sensing
Multitarget detection is very challenging, especially when targets are densely distributed. In conventional constant false alarm rate (CFAR) detection algorithms, the detection threshold is determined based on a pre-estimated background level (BL). However, interfering targets inevitably lead to inaccurate BL estimation, resulting in degraded detection performance. In our previous work, a new solution was developed and verified to be effective; in that solution, compressed sensing (CS)-based detector performs target detection without relying on BL estimation, and a subsequent CFAR regulation processor achieves the desired false alarm rate based on the statistical information of the reduced samples obtained by removing the detected targets and adjacent guard cells from the original samples. However, for scenes with a high target density, the CS-based detector suffers from performance degradation while acquiring the correlation between linear measurements of the signal and the sensing matrix due to the high local signal sparsity. To address this shortcoming, this article proposes a deep neural network (DNN)-based detector that further improves the detection performance by converting the target detection into a problem of peak sequence classification of frequency intensity (FI) measurements from radar. A DNN detector trained on augmented simulated data shows excellent generalization ability for deployment in real scenes. In addition, to achieve better computational efficiency, a Taylor-series-based approximate maximum likelihood estimator with an explicit expression is applied in the CFAR regulation process for the first time. Both simulation and field tests are performed to verify the superiority of the proposed algorithm over conventional algorithms.
17 January 2024 in IEEE Transactions on Intelligent Vehicles
Depth completion aims to recover dense depth maps from sparse depth measurements. It is a fundamental challenge in computer vision that is faced in numerous applications, such as robotics, UAVs, and autonomous driving. With the recent advancement in hardware, millimeter-wave (mmWave) radar technology is now commonly employed in high-level perception tasks of autonomous driving. However, compared to LiDAR point clouds, mmWave radar point clouds tend to be sparse and include several ghost targets. To address these challenges, we propose a semantic-guided network architecture to perform depth diffusion by utilizing the consistent relationship between semantics and depth. Our method first utilizes semantic features from images to extend the height measurements, and then transforms the global depth completion task into a series of category-specific depth diffusion tasks to learn the semantic-depth priors and accommodate radar sparsity. In addition, a gradient-guided Smooth-Edge loss is developed to explicitly constrain the semantic regional consistency and border discontinuity. Meanwhile, a 4D dataset that features Camera, 4D Radar, and LiDAR measurement in diverse scenes is collected to conduct experiments. Extensive experiments demonstrate the superior performance of our proposed method over existing monocular and fusion-based approaches.
21 March 2024 in IEEE Transactions on Intelligent Vehicles
Millimeter-wave (mmWave) radar is widely used in autonomous driving thanks to its robustness under harsh weather conditions. However, compared to LiDAR point clouds, mmWave radar point clouds tend to be sparse and include several “ghost” targets. Existing methods attempt to optimize radar signal processing to generate superior radar point cloud. Nonetheless, these methods are primarily designed for simple scenarios. How to obtain superior radar point clouds for complex autonomous driving scenarios is still unexplored. In this article, we propose a novel point cloud generation network that generates superior radar point clouds by utilizing complementary advantage of radar and camera. Technically, our approach comprises three key steps. First, to obtain precise relationships between neighboring pixels, the mmWave range-azimuth map (RAM) is enhanced to extract the radar affinity matrix. Second, to obtain dense image bird's-eye-view (BEV) features, image features are diffused by learning local details from the radar affinity matrix. Third, to obtain superior radar point clouds, radar features and diffused image features are combined in a unified BEV representation, and optimized by K-nearest-neighbor (KNN) graph. To verify the effectiveness of the proposed methods, we collect a real-world dataset under autonomous driving scenarios to conduct experiments. The results demonstrate that our method outperforms exisiting ones in producing high-quality radar point clouds. And as a consequence, the radar point cloud generated using our methods significantly improve the performance of object detection task.