Xueqing Peng, Lingfei Qian, Yan Wang, Ruoyu Xiang, Yueru He, Yang Ren, Mingyang Jiang, Vincent Jim Zhang, Yuqing Guo, Jeff Zhao, Huan He, Yi Han, Yun Feng, Yuechen Jiang, Yupeng Cao, Haohang Li, Yangyang Yu, Xiaoyu Wang, Penglei Gao, Shengyuan Lin, Keyi Wang, Shanshan Yang, Yilun Zhao, Zhiwei Liu, Peng Lu, Jerry Huang, Suyuchen Wang, Triantafillos Papadopoulos, Polydoros Giannouris, Efstathia Soufleri, Nuo Chen, Zhiyang Deng, Heming Fu, Yijia Zhao, Mingquan Lin, Meikang Qiu, Kaleb E Smith, Arman Cohan, Xiao-Yang Liu, Jimin Huang, Guojun Xiong, Alejandro Lopez-Lira, Xi Chen, Junichi Tsujii, Jian-Yun Nie, Sophia Ananiadou, Qianqian Xie. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Autonomous parking has become a critical application in automatic driving research and development. Parking operations often suffer from limited space and complex environments, requiring accurate perception and precise maneuvering. Traditional rule-based parking algorithms struggle to adapt to diverse and unpredictable conditions, while learning-based algorithms lack consistent and stable performance in various scenarios. Therefore, a hybrid approach is necessary that combines the stability of rule-based methods and the generalizability of learning-based methods. Recently, reinforcement learning (RL) based policy has shown robust capability in planning tasks. However, the simulation-to-reality (sim-to-real) transfer gap seriously blocks the real-world deployment. To address these problems, we employ a hybrid policy, consisting of a rule-based Reeds-Shepp (RS) planner and a learningbased reinforcement learning (RL) planner. A real-time LiDARbased Occupancy Grid Map (OGM) representation is adopted to bridge the sim-to-real gap, leading the hybrid policy can be applied to real-world systems seamlessly. We conducted extensive experiments both in the simulation environment and real-world scenarios, and the result demonstrates that the proposed method outperforms pure rule-based and learningbased methods. The real-world experiment further validates the feasibility and efficiency of the proposed method.

Autonomous navigation is a fundamental task for robot vacuum cleaners in indoor environments. Since their core function is to clean entire areas, robots inevitably encounter dead zones in cluttered and narrow scenarios. Existing planning methods often fail to escape due to complex environmental constraints, high-dimensional search spaces, and high difficulty maneuvers. To address these challenges, this paper proposes an embodied escaping model that leverages a reinforcement learning-based policy with an efficient action mask for dead zone escaping. To alleviate the issue of the sparse reward in training, we introduce a hybrid training policy that improves learning efficiency. In handling redundant and ineffective action options, we design a novel action representation to reshape the discrete action space with a uniform turning radius. Furthermore, we develop an action mask strategy to select valid actions quickly, balancing precision and efficiency. In real-world experiments, our robot is equipped with a Lidar, IMU, and two-wheel encoders. Extensive quantitative and qualitative experiments across varying difficulty levels demonstrate that our robot can consistently escape from challenging dead zones. Moreover, our approach significantly outperforms compared path planning and reinforcement learning methods in terms of success rate and collision avoidance. A video showcasing our methodology and real-world demonstrations is available at https://youtu.be/kBaaYWGhNuE.

The global navigation satellite system (GNSS) is commonly used for high-precision vehicle positioning by leveraging positioning satellites. However, this system faces challenges, such as occlusion and absorption. As a result, meeting the precise positioning requirements of fully autonomous vehicles becomes unattainable with GNSS alone. This article introduces a methodology that incorporates the use of signal of opportunity (SOP), inertial navigation system (INS), and map data to achieve reliable relative positioning in practical scenarios where reliance on the global navigation satellites falls short. The proposed algorithm uses opportunistic signals to correct INS errors for real-time positioning and integrates map data to accurately estimate the vehicle’s trajectory in space and time without relying on GNSS. A particle filter is employed to continuously estimate the vehicle’s position and velocity. Simulation and experimental results with ultra-wideband (UWB) SOPs are presented to evaluate the efficacy and accuracy of the proposed scheme in environments with intermittent GNSS degradation. The experimental outcomes show that the average relative positioning error of the proposed method is 0.2759 m, with the minimum and maximum relative errors being 0.0036 and 0.5446 m, respectively. Notably, the average relative positioning error is reduced by 86.9% and 31.9%, respectively, compared to GNSS/INS and UWB/INS positioning methods alone. To enhance the algorithm execution efficiency, this article also compares the average relative positioning error and execution efficiency under different particle numbers. Simulation results reveal that with 2000 particles, a tradeoff between the average relative positioning error and algorithm execution time can be achieved.

As a prominent and appealing paradigm, federated edge learning (FEEL) aims to orchestrate collaborative training across a multitude of distributed edge devices without the need for sensitive information transfer. However, device heterogeneity and wireless transmission distortion, which compromise training robustness and efficiency, severely hinder FEEL deployment in real-world applications. To this end, we propose in this paper a novel FEEL framework empowered by simultaneous transmitting and reflecting reconfigurable intelligent surface (STAR-RIS) with flexible aggregation. This framework harmonizes devices to train models with heterogeneous intensity in each communication round, while the deployment of STAR-RIS significantly boosts the quality of wireless aggregation. Here, we emphasize that there may exist a crucial trade-off between the heterogeneous local training intensity and the transmission quality, particularly when edge devices operate under a limited energy budget. To illuminate this perspective, we rigorously derive a novel explicit upper bound that captures the joint impact of local training accuracy and the mean square error of wireless aggregation on FEEL convergence performance. Our theoretical results indicate that a blind focus on improving the local training accuracy within a constrained energy budget may ultimately detract from overall training performance. This finding sharply contrasts with existing research, which typically aims to accelerate convergence by increasing local training intensity while neglecting the impact of wireless aggregation distortion. To strike the ideal balance, we formulate a mixed-integer nonlinear programming problem to guide the joint design of the beamforming at devices and the BS, the configuration mode of STAR-RIS, and the local training intensity. Comprehensive experiments on representative datasets demonstrate that our proposed framework achieves significant performance improvements compared with existing baselines.

Automated parking stands as a highly anticipated application of autonomous driving technology. However, existing path planning methodologies fall short of addressing this need due to their incapability to handle the diverse and complex parking scenarios in reality. While non-learning methods provide reliable planning results, they are vulnerable to intricate occasions, whereas learning-based ones are good at exploration but unstable in converging to feasible solutions. To leverage the strengths of both approaches, we introduce Hybrid pOlicy Path plannEr (HOPE). This novel solution integrates a reinforcement learning agent with Reeds-Shepp curves, enabling effective planning across diverse scenarios. HOPE guides the exploration of the reinforcement learning agent by applying an action mask mechanism and employs a transformer to integrate the perceived environmental information with the mask. To facilitate the training and evaluation of the proposed planner, we propose a criterion for categorizing the difficulty level of parking scenarios based on space and obstacle distribution. Experimental results demonstrate that our approach outperforms typical rule-based algorithms and traditional reinforcement learning methods, showing higher planning success rates and generalization across various scenarios. We also conduct real-world experiments to verify the practicability of HOPE. The code for our solution is openly available onhttps://github.com/jiamiya/HOPE.

As a heterogeneous disease, prostate cancer (PCa) exhibits diverse clinical and biological features, which pose significant challenges for early diagnosis and treatment. Metabolomics offers promising new approaches for early diagnosis, treatment, and prognosis of PCa. However, metabolomics data are characterized by high dimensionality, noise, variability, and small sample sizes, presenting substantial challenges for classification. Despite the wide range of applications of deep learning methods, the use of deep learning in metabolomics research has not been extensively explored. In this study, we propose a hybrid model, TransConvNet, which combines transformer and convolutional neural networks for the classification of prostate cancer metabolomics data. We introduce a 1D convolution layer for the inputs to the dot-product attention mechanism, enabling the interaction of both local and global information. Additionally, a gating mechanism is incorporated to dynamically adjust the attention weights. The features extracted by multi-head attention are further refined through 1D convolution, and a residual network is introduced to alleviate the gradient vanishing problem in the convolutional layers. We conducted comparative experiments with seven other machine learning algorithms. Through five-fold cross-validation, TransConvNet achieved an accuracy of 81.03% and an AUC of 0.89, significantly outperforming the other algorithms. Additionally, we validated TransConvNet's generalization ability through experiments on the lung cancer dataset, with the results demonstrating its robustness and adaptability to different metabolomics datasets. We also proposed the MI-RF (Mutual Information-based random forest) model, which effectively identified key biomarkers associated with prostate cancer by leveraging comprehensive feature weight coefficients. In contrast, traditional methods identified only a limited number of biomarkers. In summary, these results highlight the potential of TransConvNet and MI-RF in both classification tasks and biomarker discovery, providing valuable insights for the clinical application of prostate cancer diagnosis.

With the development of economy as well as the quality of people's life, the cars' number has increased very fast, and the condition of transportation is also becoming much more complex. Thus, it's necessary to come up with an efficient algorithm to estimate the travel time of cars. "Travel time estimation" is an important problem in intelligent transportation field, and is also pretty common in transportation monitoring and routing system. In this article, based on the work of Wang Y et al, we use data mining method to find a way to estimate the travel time of cars on any path efficiently. Firstly, considering the paths' condition, we use tensor decomposition to overcome the sparsity problem of trajectory data. Secondly we construct an evaluation function, and use dynamic programming with suffix tree optimization to choose the best combination of sub-paths. All the code are reproduced by Python language rather than Matlab, which is originally used in the work of Wang Y et al. The experiment suggests that this method can estimate travel time effectively. In this article, we investigate the study about this problem both domestically and internationally, and based on which improve the original algorithm by replacing SGD by Adam. We summarize the theory, provide detailed process of the algorithm and show the key code, after which we examine the effectiveness of it by conducting experiments. In the summary chapter, we analysis the pros and cons of the algorithm objectively, and come up with the direction of our study in the future.

We propose a low-rank subspace recovery and image denoising method for face recognition. Traditional subspace methods commonly assume that face images from a single class lie on a low-rank subspace. However, due to shadows, specularities, occlusion and corruption, real face images seldom reveal such low-rank structure. To address this problem, we cast the problem of recovering face subspace from noisy images as a problem of recovering a low-rank matrix with sparse error of arbitrary large magnitude. By using the recent breakthroughs in convex optimization, we can exactly recover the subspaces from corrupted facial data. We apply this method to two well-known subspace methods: nearest subspace and sparse representation face recognition. The results show that our method is efficient in recovering the low-rank face subspaces by removing the noise in the training images, thus significantly improve the robustness of these methods in the presence of occlusion and corruption in both train and test images.