Autonomous vehicles face significant motion planning challenges in interactive scenarios, particularly when coordinating with dynamic traffic participants. Key issues include vehicle interdependencies, dynamic priority allocation, and real-time interactions, which critically impact traffic safety and efficiency. To address these issues, we propose a Stackelberg game-theoretic motion planning framework incorporating opportunity cost—an economic concept representing the value of the best alternative strategy foregone when making a decision. Firstly, a hybrid path planner is developed to capture the interdependencies among traffic participants. Secondly, road priority during interactions is quantified through a leader-follower model. Thirdly, a Stackelberg game-based speed planner is designed to enable the autonomous vehicle to adapt to interactive environments. The payoff function, which incorporates opportunity costs via reactive speed planning, is validated using the nuPlan dataset, and its parameters are identified accordingly. The framework facilitates decision-making through exploratory planning outcomes, effectively bridging the gap between decision and planning modules. It significantly enhances driving efficiency at intersections and demonstrates strong adaptability across various interactive scenarios in both simulations and real-world experiments.

The hyperspectral image (HSI) has been widely used in many applications due to its fruitful spectral information. However, the limitation of imaging sensors has reduced its spatial resolution that causes detail loss. One solution is to fuse the low spatial resolution HSI (LR-HSI) and the panchromatic (PAN) image with inverse features to get the high-resolution HSI (HR-HSI). Most of the existing fusion methods just focus on small fusion ratios like 4 or 6, which might be impractical for some large ratios’ HSI and PAN image pairs. Moreover, the ill-posedness of restoring detail information in HSI with hundreds of bands from PAN image with only one band has not been solved effectively, especially under large fusion ratios. Therefore, a lightweight unmixing-based PAN-guided fusion network (Pgnet) is proposed to mitigate this ill-posedness and improve the fusion performance significantly. Note that the fusion process of the proposed network is under the projected low-dimensional abundance subspace with an extremely large fusion ratio of 16. Furthermore, based on the linear and nonlinear relationships between the PAN intensity and abundance, an interpretable PAN detail inject network (PDIN) is designed to inject the PAN details into the abundance feature efficiently. Comprehensive experiments on simulated and real datasets demonstrate the superiority and generality of our method over several state-of-the-art (SOTA) methods qualitatively and quantitatively (the codes in pytorch and paddle versions and dataset could be available athttps://github.com/rs-lsl/Pgnet).

Graph structural information such as topologies or connectivities provides valuable guidance for graph convolutional networks (GCNs) to learn nodes' representations. Existing GCN models that capture nodes' structural information weight in- and out-neighbors equally or differentiate in- and out-neighbors globally without considering nodes' local topologies. We observe that in- and out-neighbors contribute differently for nodes with different local topologies. To explore the directional structural information for different nodes, we propose a GCN model with weighted structural features, named WGCN. WGCN first captures nodes' structural fingerprints via a direction and degree aware Random Walk with Restart algorithm, where the walk is guided by both edge direction and nodes' in- and out-degrees. Then, the interactions between nodes' structural fingerprints are used as the weighted node structural features. To further capture nodes' high-order dependencies and graph geometry, WGCN embeds graphs into a latent space to obtain nodes' latent neighbors and geometrical relationships. Based on nodes' geometrical relationships in the latent space, WGCN differentiates latent, in-, and out-neighbors with an attention-based geometrical aggregation. Experiments on transductive node classification tasks show that WGCN outperforms the baseline models consistently by up to 17.07% in terms of accuracy on five benchmark datasets.

In this paper, we present a context-based information refinding system called ReFinder. It leverages human's natural recall characteristics and allows users to refind files and Web pages according to the previous access context. ReFinder refinds information based on a query-by-context model over a context memory snapshot, linking to the accessed information contents. Context instances in the memory snapshot are organized in a clustered and associated manner, and dynamically evolve in life cycles to mimic brain memory's decay and reinforcement phenomena. We evaluate the scalability of ReFinder on a large synthetic data set. The experimental results show that consistent degradation of context instances in the context memory and the ones in user's refinding requests can lead to the best refinding precision and recall. An 8-week user study is also conducted to examine the applicability of ReFinder. Initial findings show that time, place, and activity could serve as useful recall clues. On average, 15.53 seconds are needed to complete a refinding request with ReFinder and 84.42 seconds with other existing methods. Some further possible improvement of ReFinder is also discussed at the end of the paper.