Chao Wang

Recently, the integration of electric vehicles (EVs) into intelligent transportation systems has introduced significant and novel challenges, necessitated by the need to account for additional factors such as battery consumption and charging station characteristics. To achieve the sustainable integration of EVs into these systems, the development of new models, methods, and technologies is crucial. The electric vehicle routing problem with time windows (EVRPTW) represents an innovative challenge in this domain. It involves determining optimal routes for a fleet of EVs to provide freight transportation services, while considering energy consumption, charging times, and customer service times. Although existing methods incorporating attention-based Transformer architecture have achieved remarkable success in solving traditional vehicle routing problems, they encounter substantial difficulties when directly applied to the EVRPTW. This is mainly due to the complex interplay between customer time-window restrictions and battery-capacity limitations. To address these challenges, this paper presents a novel heuristic-augmented attention method that harnesses heuristic information to enhance attention mechanisms. Specifically, an adaptive compatibility module is introduced to leverage time windows information, thereby enhancing the effective connectivity of nodes under various constraints and improving the understanding of the solution space. Furthermore, a dynamic heuristic-enhanced attention is utilized to facilitate a more in-depth understanding of the relationships between different types of nodes. The experimental results demonstrate that the proposed approach out performs representative attention-based methods. It achieves superior performance in balancing route length and time window satisfaction.

The electric vehicle routing problem (EVRP) is a complex logistical challenge that requires optimizing both route planning and charging decisions. Although ant colony optimization (ACO) has demonstrated potential in solving the EVRP, conventional approaches often rely on simplistic distance-based heuristics that inadequately capture the problem’s global characteristics, resulting in low search efficiency and limited generalization capability. To overcome these limitations, we propose a novel ACO with deep heuristic information (ACO-DHI), where a learning model is designed to extract richer heuristic information from a large number of offline instances. This model incorporates a heterogeneous attention mechanism to capture the intrinsic relationships between customers and charging stations, and a feature refinement method to gain more relevant problem-specific information. Once offline training is completed, the model outputs a deep heuristic information matrix, which replaces the traditional distance matrix in ACO for online optimization. Additionally, a new pheromone updating strategy is designed specifically for EVRP, adaptively updating pheromone levels in response to different visits to customers and charging stations. Extensive computational experiments on public benchmark instances demonstrate that ACO-DHI exhibits faster solution speed and stronger generalization capability compared to existing learning methods and heuristics. Code is available at https://github.com/ACO-DHI/ACO-DHI

The Dynamic Vehicle Routing Problem with Time Windows (DVRPTW) is a widespread real-world challenge, and numerous algorithms have been proposed to address it. However, in the context of an emerging logistics paradigm, namely the instant delivery, the performance of existing algorithms tailored for DVRPTW degrades significantly, as instant delivery allows only very limited computational time for solving DVRPTW instances. Owing to the periodic nature of customer orders, this paper proposes a data-driven evolutionary algorithm (DDEA) for solving DVRPTW under limited computation time. In the offline phase, a set of generalized solutions is derived from historical data via a dedicated evolutionary algorithm. These solutions are then directly employed in the online phase to construct high-quality solutions for new problem instances. By leveraging these precomputed generalized solutions, DDEA effectively operates within tight time constraints. Extensive experiments using synthetic and real-world data demonstrate that DDEA outperforms five state-of-the-art algorithms designed for DVRPTW under limited computation time, particularly under extremely short time constraints.

The Multi-depot Vehicle Routing Problem with Uncertain Demand (MD-VRPUD) can be modeled as a bi-level optimization problem (BLOP), because it requires both assigning customers to different depots and determining the routes for servicing customers, where the optimization of these two parts is coupled with each other. Although the bi-level evolutionary algorithm is a fitting approach for tackling the MD-VRPUD, its nested structure often leads to computational inefficiency. To this end, this paper tailors a surrogate-assisted bi-level evolutionary algorithm (SABLEA) to achieve highly efficient nested algorithms tailored for solving the MD-VRPUD. To deal with the combinatorial property of MD-VRPUD, two groups of continuous features are first extracted to help the surrogate model to effectively distinguish the superiority and inferiority of schemes. Then, a management strategy is designed to adaptively build the surrogate models in different subspaces so as to alleviate the performance bottleneck faced by the model. Finally, an adaptive computing resource allocation strategy is integrated into the lower-level optimization, to allocate more resources to promising customer assignment schemes, enabling the discovery of better routes and improving the overall accuracy of models. The comprehensive experimental results demonstrate the effectiveness of the SABLEA in handling MD-VRPUD, outperforming four existing algorithms in terms of both computational efficiency and solution quality.

Recently, the electric vehicle routing problem (EVRP) has emerged as a significant challenge in the transportation industry. Unlike traditional vehicle routing problems (VRPs) that focus primarily on route optimization for customers, the EVRP must also consider electricity constraints and charging decisions. Although deep reinforcement learning (DRL), particularly with attention-based Transformer architecture, has achieved success in VRPs, it faces significant difficulties when directly applied to the EVRP due to the additional multifaceted constraints and node types. To address these challenges, this paper proposes an attention-enhanced DRL method specifically designed for the EVRP. First, a heterogeneous attention mechanism is developed in the encoder to capture the intrinsic access relationships between customers and charging stations. Then, an embedding enhancement module is integrated to enrich node embeddings by incorporating adjacent distance and solution structure information through a global-local attention mechanism. Finally, a problem-size sensor is incorporated within the decoder to quantify route planning progress, thereby enhancing adaptability to the EVRP constraints. Empirical results demonstrate that the proposed method not only achieves higher efficiency compared to heuristic methods, but also significantly outperforms representative DRL-based approaches in terms of solution quality.

Stereo matching for inland waterways is one of the key technologies for the autonomous navigation of Unmanned Surface Vehicles (USVs), which involves dividing the stereo images into reference images and target images for pixel-level matching. However, due to the challenges of the inland waterway environment, such as blurred textures, large spatial scales, and computational resource constraints of the USVs platform, the participation of geometric features from the target image is required for efficient target-driven matching. Based on this target-driven concept, we propose a lightweight target-driven stereo matching neural network, named LTNet. Specifically, a lightweight and efficient 4D cost volume, named the Geometry Target Volume (GTV), is designed to fully utilize the geometric information of target features by employing the shifted target features as the filtered feature volume. Subsequently, to address the substantial texture interference and object occlusions present in the waterway environment, a Left-Right Consistency Refinement (LRR) module is proposed. The \text{LRR} utilizes the pixel-level differences in left and right disparities to introduce soft constraints, thereby enhancing the accuracy of predictions during the intermediate stages of the network. Moreover, knowledge distillation is utilized to enhance the generalization capability of lightweight models on the USVInland dataset. Furthermore, a new large-scale benchmark, named Spring, is utilized to validate the applicability of LTNet across various scenarios. In experiments on the aforementioned two datasets, LTNet achieves competitive results, with only 3.7M parameters. The code is available at https://github.com/Open-YiQingZhou/LTNet .