The development of vehicles-to-everything (V2X) communication and autonomous driving technologies offers novel opportunities for eco-driving of connected hybrid electric vehicles (HEVs). To enhance vehicle energy efficiency in complex traffic scenarios, this study proposes a novel interaction-aware eco-driving strategy utilizing a Safe Soft Actor-Critic (Safe-SAC) deep reinforcement learning (DRL) algorithm for connected HEVs, which jointly optimizes speed planning and lane-changing decisions. By leveraging V2X technology, multi-source traffic information is integrated into the DRL environment, where the state space encompasses the states of the ego vehicle, traffic flow, signal phase and timing (SPaT), and the states of surrounding vehicles, while the action space contains the target lane and desired acceleration of the ego vehicle. In contrast to previous works, the integration of traffic flow speed and lane occupancy for each lane as the state variables enables a more accurate representation of the dynamic characteristics of surrounding traffic. Furthermore, safety constraints and a multi-objective reward function are meticulously designed to balance energy efficiency, driving comfort, and travel efficiency while ensuring safety. To thoroughly evaluate the energy efficiency of the proposed eco-driving strategy, both dynamic programming (DP) and equivalent consumption minimization strategy (ECMS) are employed as the underlying energy management strategies (EMS). Finally, the effectiveness of the Safe-SAC strategy is successfully validated on the co-simulation platform based on the Simulation of Urban Mobility (SUMO) and Python under various traffic scenarios. Compared to the Krauss-LC2013 model, the findings highlight the superiority of the proposed Safe-SAC strategy in achieving an average energy saving of 38.5%. This strategy also enhances driving comfort and maintains higher travel efficiency.

The craters widely distributed on the lunar surface constitute important landforms for investigating the geological activities and evolutionary history of the Moon. Currently, deep learning models have made certain progress in detecting craters, but they still have limitations in identifying numerous small craters. To address these challenges, this article puts forward a CenterNet-based dense small crater detection model (DSCraterNet), which is designed to precisely detect craters within the permanently shadowed regions (PSRs) by leveraging the previously released enhanced PSR images. DSCraterNet achieves precise crater localization by predicting keypoint heatmap (location), size, and offset of craters. The model employs an encoder–decoder network architecture that combines various feature fusion and attention mechanisms. Feature fusion strategies include direct fusion between the encoder and decoder, multiscale encoder spatial feature fusion (MSF) structure, and multielement information fusion (MEF) structure. These fusion strategies balance the relationship between high-resolution features, strong semantic features, and multiscale spatial features according to craters’ characteristics. Furthermore, by reasonably configuring attention-based layers [like efficient multiscale attention (EMA)] at different stages, the utilization efficiency of robust features is improved. Ultimately, DSCraterNet successfully identified a large number of densely packed small craters in PSR, revealing the terrain features of the region. A series of thorough ablation and comparative experiments has confirmed that DSCraterNet surpasses existing crater detection models in terms of precision, recall, mean average precision (mAP), and$F1$-score metrics. Our code is available athttps://github.com/dl-zfq/DSCraterNet

Traffic congestion frequently occurs on slope segments of highways, which is a typical bottleneck. With the development of connected and autonomous vehicle (CAV) technology, there will be a scene that CAVs and human driven vehicles (HDVs) co-exist. This work studies a slope bottleneck on highway in mixed traffic scenarios and proposes a traffic flow model for slope bottlenecks incorporating CAV platooning based on cellular automata. A novel traffic flow control strategy for slope bottlenecks is proposed based on variable speed limit (VSL) and vehicle platooning. Firstly, it divides the upstream section of the slope bottleneck into two zones for implementing VSL and vehicle platooning, respectively. Via speed restrictions within the VSL zone, the inflow of vehicles into the vehicle platooning zone is effectively mitigated to create low traffic density. In the vehicle platooning zone, a hybrid vehicle platooning method for mixed scenarios is proposed. The experimental results demonstrate that our strategy effectively enhances traffic flow of the slope bottleneck, thereby mitigating traffic congestion.

Obtaining high-visibility images of the lunar polar permanently shadowed region (PSR) is quite important for internal landforms and material existence exploration. However, PSR images usually have poor quality due to a lack of sufficient illumination. Existing researches, that attempt to address this problem, face challenges caused by relying on virtual assumptions, manual processing, and paired data. To solve these problems, we aim to avoid using paired datasets and directly optimize PSR images, and accordingly propose a zero-shot parameter learning model (ZSPL-PSR) for PSR image enhancement. Our ZSPL-PSR, which enhances PSR images by estimating parameters to adjust image properties, consists of a parameter learning network and a parameter weight learning structure. Particularly, first, a parameter learning network that integrates robust information is constructed to separately estimate the midtone brightness parameters, shadow brightness parameters, and contrast parameters. Where these parameters are beneficial for iteratively improve the overall brightness, shadow brightness, and contrast of the image. Second, a parameter weight learning structure is exploited to coordinate the priority of different parameter maps. In addition, to highlight the terrain details in the enhanced PSR image, we use USM sharpening for postprocessing. The experimental results display the fully interpretable enhanced PSR maps of the lunar north and south poles and their sharpened versions, showcasing rich landforms in PSR. To validate the model performance, a benchmark PSR testing set has been constructed, and extensive comparisons conducted on it demonstrated that ZSPL-PSR exceeds other zero-shot learning methods significantly in image quality. Our code is available athttps://github.com/dl-zfq/ZSPL-PSR.

Connected and automated vehicle technology via vehicle-to-everything communication, can assist in improving energy efficiency for hybrid electric vehicles (HEVs). In particular, information about the timing of traffic lights and surrounding vehicles can be exchanged between traffic vehicles and in conjunction with vehicle state information, to improve the fuel economy of HEVs significantly. To this end, we propose a multi-lane hierarchical optimization (MLHO) algorithm based on a predictive control framework. The dynamic behaviors of the surrounding vehicles are first predicted, and then the traffic light information (e.g., signal phasing and timing) and vehicles’ state information are utilized in the design. MLHO is a two-level strategy wherein a multi-lane speed planning method for a host vehicle is formulated to plan the optimal speed and lane-change behaviors by considering vehicle power demand, driving comfort, and safety in the upper level. In the lower level, dynamic programming is adopted to devise energy management by tracking the optimal speed. Simulation results under real routes using the traffic simulation software Simulation of Urban Mobility show that the fuel economy of MLHO is improved by 32% on average compared to speed profile driven by a human driver model. In addition, traffic efficiency is enhanced significantly, i.e., different traffic occupancy results on the road indicate that the proposed MLHO is less affected by the traffic flow density. With different traffic densities, the maximum fuel consumption difference under the three considered scenarios is only 0.645L/100km.

Reliable and accurate velocity prediction can significantly contribute to the quality of connected vehicle control applications. Existing efforts focus on the velocity prediction without considering vehicle-to-vehicle (V2V) communication interruption. Hence, a stochastic velocity prediction method for connected vehicles considering V2V communication interruption is put forward for the first time. The missing V2V communication data are addressed by the piecewise cubic Hermite spline interpolation. Then, the processed data are used as the input variables of the best conditional linear Gaussian (CLG) prediction model. Specifically, the best CLG model is obtained by analyzing the influence of different input variables on the velocity prediction without V2V communication interruption. The results demonstrate that the prediction accuracy of CLG-based model is acceptable if the communication interruption time is less than 5 s compared to the non-interrupted V2V communication case. The sensitivity study of the best CLG model under multiple vehicles scenario indicates that choosing appropriate historical data substantially improve the prediction accuracy. Furthermore, the CLG-based predictor is proved to be an effective method to achieve higher prediction accuracy in two test road networks when compared with the Back-propagation and Long Short-Term Memory network.

Simultaneous extraction of spectral and spatial features and their fusion is currently a popular solution in hyperspectral image (HSI) classification. It has achieved satisfactory results in some research. Because the scales of objects are often different in HSI, it is necessary to extract multi-scale features. However, this aspect was not taken into account in many spectral-spatial feature fusion methods. This causes the model to be unable to get sufficient features on scales with a large difference range. The model (MCMN: Multi-Content Merging Network) proposed in this paper designs a multi-branch fusion structure to extract multi-scale spatial features by using multiple dilated convolution kernels. Considering the interference of the surrounding heterogeneous objects, the useful information from different directions is also fused together to realize the merging of multiple regional features. MCMN introduces a convolution block attention mechanism, which fully extracts attention features in both spatial and spectral directions, so that the network can focus on more useful parts, which can effectively improve the performance of the model. In addition, since the number of objects in each class is often discrepant, it will have some impact on the training process. We apply the focal loss function to eliminate the negative factor. The experimental results of MCMN on three data sets have a breakthrough compared with the other comparison models, which highlights the role of MCMN structure.

The performance of energy management in hybrid electric vehicles is highly dependent on the forecasted velocity. To this end, a new velocity-prediction approach utilizing the concept of chaining neural network (CNN) is introduced. This velocity forecasting approach is subsequently used as the basis for an equivalent consumption minimization strategy (ECMS). The CNN is used to predict the velocity over different temporal horizons, exploiting the information provided through vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication channels. In addition, a new adaptation law for the so-called equivalent factor (EF) in ECMS is devised to investigate the effects of future velocity on fuel economy and to impose charge sustainability. Compared with traditional adaptation law, this paper considers the impact of predicted velocity on EF. The control objective is to improve the fuel economy relative to the ECMS without considering predicted velocity. Finally, simulations are conducted in three cases over different prediction horizons to demonstrate the performance of the proposed velocity-prediction method and ECMS with adaptation law. Simulation results confirm that ECMS with EF adjusted by the proposed adaptation law produces between 0.2% and 5% improvements in fuel economy relative to ECMS with traditional adaptation law. In addition, better charge sustainability is achieved as well.

This paper proposes a new energy management based on equivalent consumption minimization strategy (ECMS) for hybrid electric vehicles. The aim is to impose SoC charge-sustainability and enhance the fuel economy. First, the equivalent factor (EF) of ECMS is derived from Pontryagin's Minimum Principle. Second, a new adaptation law using Fuzzy Proportional plus Integral (PI) controller is developed to adjust EF in real-time. Finally, simulations for two driving cycles using ECMS are compared with rule-based (RB) control strategy, indicating that the proposed adaptation law can provide a promising blend in terms of fuel economy and charge-sustainability. The results show that ECMS with Fuzzy PI adaptation of EF achieves significant improvement compared with RB in terms of fuel economy and is more robust than ECMS with constant EF.