Prototype-based personalized federated learning methods have emerged as a promising strategy due to their ability to represent client-specific class characteristics effectively through learned class prototypes. These prototypes capture salient features of client-local data, facilitating personalized model adaptation. However, existing prototype-based aggregation strategies predominantly rely on weighted averaging, implicitly assuming prototype consistency across clients. This assumption neglects the intrinsic heterogeneity and non-independent and identically distributed (non-IID) nature of client data, compelling diverse local prototypes to align toward a singular global prototype and consequently causing significant aggregation bias. Motivated by observations from intra-class feature saliency analysis, we identify that clients inherently emphasize distinct feature regions even for the same class. To leverage this intra-class diversity, we introduce FedIC, a novel prototype clustering and collaborative classifier optimization approach. Specifically, FedIC first clusters prototypes based on intra-class similarity to form intra-class prototype subspaces, ensuring that aggregation occurs exclusively within each cluster, thus eliminating the bias stemming from forced global unification. To further exploit the benefits of intra-cluster collaboration, we quantify the combined predictive gains of classifiers from clients within the same cluster as a function of classifier combination weights. This targeted aggregation and collaborative optimization strategy effectively circumvents the bias introduced by global alignment. Extensive experiments under various non-IID settings show that FedIC significantly outperforms existing Prototype-based and Clustered PFL Methods.

Movable antennas (MAs) fully exploit spatial degrees of freedom (DoFs) in continuous space through dynamic position adjustment, thereby enhancing the performance of integrated sensing and communication (ISAC) systems. This paper investigates robust beamforming and antenna position optimization in MA assisted ISAC systems under bounded channel uncertainty. A position-dependent channel estimation error model is first derived, jointly accounting for errors induced by antenna movement, angle estimation, and signal strength estimation. Based on this model, a robust optimization problem is formulated to maximize sensing performance while satisfying the signal-to-interference-plus-noise ratio (SINR) requirements of multiple communication users. To address the non-convexity arising from the coupling between antenna positions and beamforming vectors, a low-complexity alternating optimization algorithm is developed, incorporating the S-procedure, semidefinite relaxation (SDR), and successive convex approximation (SCA). Simulation results demonstrate that the proposed algorithm significantly outperforms fixed-position antenna (FPA)-based schemes, achieving up to a 40% improvement in sensing beampattern gain while ensuring the SINR requirements.

Accurate multi-label classification of thoracic diseases from chest X-rays continues to be challenging due to strong inter-class correlations and limited generalization of traditional convolutional approaches. In this work, we propose a Causal-Aware Swin Transformer (CAST) framework that integrates causal inference components with Masked Autoencoder (MAE) pretraining, for better representation learning on the NIH ChestX-ray14 dataset. In particular, we develop a Causal Attention module that explicitly learns the causal contribution of latent features while contrasting original and counterfactual representations, allowing the network to focus on cues related to the disease while suppressing spurious correlations. The Swin Transformer backbone, initialized with MAE pretraining, also enhances the model's ability to learn hierarchically from limited supervision. Our empirical evidence shows improved classification performance in 14 categories of thoracic disease, achieving higher mean AUC than existing state-of-the-art methods. This proposed framework highlights the effectiveness of combining causal reasoning with transformer based methods as a promising approach for interpretable and robust multi-label modeling for medical image classification.

Next Point-of-Interest (POI) recommendation, which predicts users' future destinations based on their potential interests, has emerged as a critical task in location-based social networks (LBSNs). However, this task remains challenged by issues such as popularity bias, exposure bias, and limited representational capacity, all of which impede the accurate modeling of users and POIs, thereby restricting balanced and effective recommendations. Therefore, we propose Trajectory Collaborative Filtering Mamba (TCFMamba), which integrates two specially designed modules, i.e., Joint Learning of Static and Dynamic Representations (JLSDR) and Preference State Mamba Network (PSMN), for debiased Point-of-Interest recommendation.

Many existing personalized federated learning (PFL) methods utilize clustering-based aggregation to group clients with similar data characteristics, improving model performance by promoting collaboration among clients with shared features. While this method effectively mitigates some challenges posed by data heterogeneity, it predominantly relies on static data features, making it challenging to capture the dynamic changes in client models during iterative training. This limitation impedes accurate clustering based on evolving model updates. To address this issue, we propose a Gradient-Driven Adaptive Clustering method in PFL (GradPFL), which more effectively captures the personalized deviations in locally updated models. Our approach also introduces an adaptive historical gradient mechanism that refines the clustering process by incorporating both current and past update characteristics. This enables more accurate model aggregation that adapts to ongoing changes in client models during training. Experimental results demonstrate that GradPFL outperforms existing clustering-based PFL methods, especially in more complex non-IID environments.

Synthetic Aperture Radar (SAR) has the ability of all-weather and all-day observation, which is suitable for flood monitoring. However, there still has some challenges in flood monitoring using SAR images, such as high-quality prior knowledge, accumulated errors in time series analysis, and the impact of other land cover changes. To solve these problems, in this paper, a novel unsupervised flood monitoring method using temporal information and statistical characteristics of SAR images is proposed. Firstly, the temporal-spatial-polarization (TSP) dataset is constructed from different SAR data. Furthermore, improved K-means clustering is proposed to fit these constructed datasets to mitigate the error accumulation. Finally, considering the statistical characteristics of SAR data, the Bray-Curtis distance is applied to optimize improved K-means. To verify the effectiveness of the proposed method, the latest flood event in Jingpo Lake in China with temporal Sentinel-1 data is used. The experimental results demonstrate that our method has superior performance in detecting flood regions, with OA and Kappa of up to 97.64% and 0.86.

The key to friendly collaboration in vehicular communication systems lies in the reliable communication between vehicles. The current systems, which employ fixed transmission schemes, significantly constrain system capacity in respect of spectrum. Besides, considering the dynamic environment of vehicular communications, the requirement for real-time scenario identification is particularly urgent. Hence, in order to improve the reliability of communications, this paper proposes an adaptive modulation and coding (AMC) technique driven by deep learning (DL)-based scenario identification (SI) in vehicular communication systems, namely SI-AMC. In contrast to the traditional AMC technique, our proposed SI-AMC attains a refined channel response estimation through the pre-discrimination of scenario features, thereby further enhancing vehicular communication performance. During the transmission process, the SI-AMC scheme achieves environment adaptability through rate-adaptive adjustments. Moreover, in terms of SI, we creatively design an enhanced convolutional neural network structure which ex-ploits a novel activation function and regularization strategies to enhance the robustness of the model. Simulation results show that the accuracy of SI reaches up to 97.86 % and the throughput of the entire link in vehicular communications is effectively promoted. Code and the model will be released at https://github.com/communicationDLl/SI-AMC.

In simultaneous transmit and receive wireless communications, digital self-interference (SI) cancellation is required before estimating the remote transmission (RT) channel. Considering the inherent connection between SI channel reconstruction and RT channel estimation, we propose a multi-layered M-estimate total least mean squares (m-MTLS) joint estimator to estimate both channels. In each layer, our proposed m-MTLS estimator first employs an M-estimate total least mean squares (MTLS) algorithm to eliminate residual SI from the received signal and give a new estimation of the RT channel. Then, it gives the final RT channel estimation based on the average of the estimation values obtained from each layer. Compared to traditional minimum mean square error estimator and single-layered MTLS estimator, it demonstrates that the m-MTLS estimator has better performance of normalized mean squared difference. Besides, the simulation results also show the robustness of m-MTLS estimator even in scenarios where the local reference signal is contaminated with noise, and the received signal is impacted by strong impulse noise.

Up-to-date High-Definition (HD) maps are essential for self-driving cars. To achieve constantly updated HD maps, we present a deep neural network (DNN), Diff-Net, to detect changes in them. Compared to traditional methods based on object detectors, the essential design in our work is a parallel feature difference calculation structure that infers map changes by comparing features extracted from the camera and rasterized images. To generate these rasterized images, we project map elements onto images in the camera view, yielding meaningful map representations that can be consumed by a DNN accordingly. As we formulate the change detection task as an object detection problem, we leverage the anchor-based structure that predicts bounding boxes with different change status categories. To the best of our knowledge, the proposed method is the first end-to-end network that tackles the high-definition map change detection task, yielding a single stage solution. Furthermore, rather than relying on single frame input, we introduce a spatio-temporal fusion module that fuses features from history frames into the current, thus improving the overall performance. Finally, we comprehensively validate our method's effectiveness using freshly collected datasets. Results demonstrate that our Diff-Net achieves better performance than the baseline methods and is ready to be integrated into a map production pipeline maintaining an up-to-date HD map.

We present a learning-based planner that aims to robustly drive a vehicle by mimicking human drivers’ driving behavior. We leverage a mid-to-mid approach that allows us to manipulate the input to our imitation learning network freely. With that in mind, we propose a novel feedback synthesizer for data augmentation. It allows our agent to gain more driving experience in various previously unseen environments that are likely to encounter, thus improving overall performance. This is in contrast to prior works that rely purely on random synthesizers. Furthermore, rather than completely commit to imitating, we introduce task losses that penalize undesirable behaviors, such as collision, off-road, and so on. Unlike prior works, this is done by introducing a differentiable vehicle rasterizer that directly converts the waypoints output by the network into images. This effectively avoids the usage of heavyweight ConvLSTM networks, therefore, yields a faster model inference time. About the network architecture, we exploit an attention mechanism that allows the network to reason critical objects in the scene and produce better interpretable attention heatmaps. To further enhance the safety and robustness of the network, we add an optional optimization-based post-processing planner improving the driving comfort. We comprehensively validate our method’s effectiveness in different scenarios that are specifically created for evaluating self-driving vehicles. Results demonstrate that our learning-based planner achieves high intelligence and can handle complex situations. Detailed ablation and visualization analysis are included to further demonstrate each of our proposed modules’ effectiveness in our method.

Environmental fluctuations pose crucial challenges to a localization system in autonomous driving. We present a robust LiDAR localization system that maintains its kinematic estimation in changing urban scenarios by using a dead reckoning solution implemented through a LiDAR inertial odometry. Our localization framework jointly uses information from complementary modalities such as global matching and LiDAR inertial odometry to achieve accurate and smooth localization estimation. To improve the performance of the LiDAR odometry, we incorporate inertial and LiDAR intensity cues into an occupancy grid based LiDAR odometry to enhance frame-to-frame motion and matching estimation. Multi-resolution occupancy grid is implemented yielding a coarse-to-fine approach to balance the odometry's precision and computational requirement. To fuse both the odometry and global matching results, we formulate a MAP estimation problem in a pose graph fusion framework that can be efficiently solved. An effective environmental change detection method is proposed that allows us to know exactly when and what portion of the map requires an update. We comprehensively validate the effectiveness of the proposed approaches using both the Apollo-SouthBay dataset and our internal dataset. The results confirm that our efforts lead to a more robust and accurate localization system, especially in dynamically changing urban scenarios.

Different from previous data-driven methods for lithium-ion battery State-of-Charge (SoC) estimation, this paper aims to develop a hierarchical SoC estimation method to address the data dependency issue and measurement noise interferences. In the off-line training layer, aging-aware features are extracted to improve SoC estimation accuracy throughout the entire battery life cycle. Extreme gradient boosting (XGBoost) is introduced to map the relationship between the extracted features and SoC for its strong nonlinear fitting ability. In the on-line estimation layer, Ampere-hour integral method is utilized to provide SoC reference to guarantee the stability of the proposed method. Meanwhile, to suppress the measurement noise, we adopt Kalman filter to correct the SoC value estimated by XGBoost. The superiority of the proposed method is proved under the random walk discharging experiment by comparing with the results of XGBoost, i.e., without Kalman filter. The proposed method improved the accuracy of lithium-ion battery SoC by 4% to 10%.

We present L3-Net - a novel learning-based LiDAR localization system that achieves centimeter-level localization accuracy, comparable to prior state-of-the-art systems with hand-crafted pipelines. Rather than relying on these hand-crafted modules, we innovatively implement the use of various deep neural network structures to establish a learning-based approach. L3-Net learns local descriptors specifically optimized for matching in different real-world driving scenarios. 3D convolutions over a cost volume built in the solution space significantly boosts the localization accuracy. RNNs are demonstrated to be effective in modeling the vehicle's dynamics, yielding better temporal smoothness and accuracy. We comprehensively validate the effectiveness of our approach using freshly collected datasets. Multiple trials of repetitive data collection over the same road and areas make our dataset ideal for testing localization systems. The SunnyvaleBigLoop sequences, with a year's time interval between the collected mapping and testing data, made it quite challenging, but the low localization error of our method in these datasets demonstrates its maturity for real industrial implementation.

We present DeepVCP - a novel end-to-end learning-based 3D point cloud registration framework that achieves comparable registration accuracy to prior state-of-the-art geometric methods. Different from other keypoint based methods where a RANSAC procedure is usually needed, we implement the use of various deep neural network structures to establish an end-to-end trainable network. Our keypoint detector is trained through this end-to-end structure and enables the system to avoid the interference of dynamic objects, leverages the help of sufficiently salient features on stationary objects, and as a result, achieves high robustness. Rather than searching the corresponding points among existing points, the key contribution is that we innovatively generate them based on learned matching probabilities among a group of candidates, which can boost the registration accuracy. We comprehensively validate the effectiveness of our approach using both the KITTI dataset and the Apollo-SouthBay dataset. Results demonstrate that our method achieves comparable registration accuracy and runtime efficiency to the state-of-the-art geometry-based methods, but with higher robustness to inaccurate initial poses. Detailed ablation and visualization analysis are included to further illustrate the behavior and insights of our network. The low registration error and high robustness of our method make it attractive to the substantial applications relying on the point cloud registration task.

We present a robust and precise localization system that achieves centimeter-level localization accuracy in disparate city scenes. Our system adaptively uses information from complementary sensors such as GNSS, LiDAR, and IMU to achieve high localization accuracy and resilience in challenging scenes, such as urban downtown, highways, and tunnels. Rather than relying only on LiDAR intensity or 3D geometry, we make innovative use of LiDAR intensity and altitude cues to significantly improve localization system accuracy and robustness. Our GNSS RTK module utilizes the help of the multi-sensor fusion framework and achieves a better ambiguity resolution success rate. An error-state Kalman filter is applied to fuse the localization measurements from different sources with novel uncertainty estimation. We validate, in detail, the effectiveness of our approaches, achieving 5-10cm RMS accuracy and outperforming previous state-of-the-art systems. Importantly, our system, while deployed in a large autonomous driving fleet, made our vehicles fully autonomous in crowded city streets despite road construction that occurred from time to time. A dataset including more than 60 km real traffic driving in various urban roads is used to comprehensively test our system.

We present a real-time monocular visual odometry system that achieves high accuracy in real-world autonomous driving applications. First, we demonstrate robust monocular SFM that exploits multithreading to handle driving scenes with large motions and rapidly changing imagery. To correct for scale drift, we use known height of the camera from the ground plane. Our second contribution is a novel data-driven mechanism for cue combination that allows highly accurate ground plane estimation by adapting observation covariances of multiple cues, such as sparse feature matching and dense inter-frame stereo, based on their relative confidences inferred from visual data on a per-frame basis. Finally, we demonstrate extensive benchmark performance and comparisons on the challenging KITTI dataset, achieving accuracy comparable to stereo and exceeding prior monocular systems. Our SFM system is optimized to output pose within 50 ms in the worst case, while average case operation is over 30 fps. Our framework also significantly boosts the accuracy of applications like object localization that rely on the ground plane.

We present a system for fast and highly accurate 3D localization of objects like cars in autonomous driving applications, using a single camera. Our localization framework jointly uses information from complementary modalities such as structure from motion (SFM) and object detection to achieve high localization accuracy in both near and far fields. This is in contrast to prior works that rely purely on detector outputs, or motion segmentation based on sparse feature tracks. Rather than completely commit to tracklets generated by a 2D tracker, we make novel use of raw detection scores to allow our 3D bounding boxes to adapt to better quality 3D cues. To extract SFM cues, we demonstrate the advantages of dense tracking over sparse mechanisms in autonomous driving scenarios. In contrast to complex scene understanding, our formulation for 3D localization is efficient and can be regarded as an extension of sparse bundle adjustment to incorporate object detection cues. Experiments on the KITTI dataset show the efficacy of our cues, as well as the accuracy and robustness of our 3D object localization relative to ground truth and prior works.

Scale drift is a crucial challenge for monocular autonomous driving to emulate the performance of stereo. This paper presents a real-time monocular SFM system that corrects for scale drift using a novel cue combination framework for ground plane estimation, yielding accuracy comparable to stereo over long driving sequences. Our ground plane estimation uses multiple cues like sparse features, dense inter-frame stereo and (when applicable) object detection. A data-driven mechanism is proposed to learn models from training data that relate observation covariances for each cue to error behavior of its underlying variables. During testing, this allows per-frame adaptation of observation covariances based on relative confidences inferred from visual data. Our framework significantly boosts not only the accuracy of monocular self-localization, but also that of applications like object localization that rely on the ground plane. Experiments on the KITTI dataset demonstrate the accuracy of our ground plane estimation, monocular SFM and object localization relative to ground truth, with detailed comparisons to prior art.

We present a real-time, accurate, large-scale monocular visual odometry system for real-world autonomous outdoor driving applications. The key contributions of our work are a series of architectural innovations that address the challenge of robust multithreading even for scenes with large motions and rapidly changing imagery. Our design is extensible for three or more parallel CPU threads. The system uses 3D-2D correspondences for robust pose estimation across all threads, followed by local bundle adjustment in the primary thread. In contrast to prior work, epipolar search operates in parallel in other threads to generate new 3D points at every frame. This significantly boosts robustness and accuracy, since only extensively validated 3D points with long tracks are inserted at keyframes. Fast-moving vehicles also necessitate immediate global bundle adjustment, which is triggered by our novel keyframe design in parallel with pose estimation in a thread-safe architecture. To handle inevitable tracking failures, a recovery method is provided. Scale drift is corrected only occasionally, using a novel mechanism that detects (rather than assumes) local planarity of the road by combining information from triangulated 3D points and the inter-image planar homography. Our system is optimized to output pose within 50 ms in the worst case, while average case operation is over 30 fps. Evaluations are presented on the challenging KITTI dataset for autonomous driving, where we achieve better rotation and translation accuracy than other state-of-the-art systems.