Author

Weien Zhou

View on Semantic Scholar

2 papers indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

Jul 2026

Research on Explicit Topology Construction and Shape Optimization via Hole Deformation

Topology optimization provides innovative solutions for lightweight structural design by rationally arranging material distribution. It enhances structural performance while reducing material consumption and structural weight, thereby significantly lowering production and operational costs and generating enormous economic benefits. In the development of topology optimization, the density-based method has gained widespread adoption due to its easy-to-understand principles. However, this method still faces the following challenges when applied to engineering applications. First, the geometric models generated by topology optimization lack explicit parameter descriptions, leading to data interaction barriers with Computer Aided Design (CAD) systems. Second, due to element discretization and density penalty mechanisms, structural boundaries exhibit rough and blurred characteristics. These problems severely constrain the iterative efficiency of structural design and manufacturing feasibility. To address these issues, this paper proposes a strategy for geometric reconstruction and shape optimization of topology optimization results. The reconstruction process begins with extracting isolines from the density field as a set of contour points. These points are subsequently interpolated with B-spline curves to explicitly represent the geometric boundaries. Shape optimization is then carried out by adjusting the positions of the B-spline control points. Compared to post-processing methods based on graphics techniques for topology optimization, which ignore the volume constraint and performance loss, the structures reconstructed in this paper exhibits the following advantages: structural boundaries are smoothed and characterized with explicit parameters, reducing performance loss caused by geometric reconstruction while satisfying volume constraints. This paper successfully establishes compatibility between topology optimization and CAD systems, facilitating the transition from conceptual design to manufacturing.

Yuting Tang, Yu Li, Jiaxiang Luo et al. · SAE technical paper series · 0 citations

Preprint Aug 2026

CoNav-UAV: Cooperative Dual-Altitude Aerial Navigation via Stackelberg Learning

Target-oriented vision-and-language navigation (VLN) on aerial platforms is attracting growing attention for missions such as disaster rescue, infrastructure inspection, and security patrol. In this task, an unmanned aerial vehicle (UAV) needs to locate targets given only a concise description of their appearance and surroundings. This requires global exploration and grounding as well as collision-free close-range approach, two interleaved processes difficult to reconcile within a single agent. Most existing methods transfer the ground VLN paradigm to a low-altitude UAV and compensate for its inefficient exploration with external assistance. A recent attempt deploys two UAVs at complementary altitudes yet still relies on privileged information and trains its two agents independently, precluding any mutual adaptation essential for cooperation. Here we propose CoNav-UAV, which explicitly models the task as a Stackelberg game between a high-altitude leader and a low-altitude follower, with the system operating on onboard visual and linguistic inputs alone. To solve this game, we introduce Iterative Stackelberg Learning. The leader's high-level vision-language reasoning is refined via memory-based in-context learning, while the follower's precise motion control is updated via DAgger-style expert distillation. The alternation drives both agents toward a Stackelberg equilibrium. CoNav-UAV consistently outperforms single- and dual-agent baselines across three high-fidelity urban scenes from the AerialVLN benchmark. Success rate improves by up to 30.8 points on the learning scene, and 9.0 points under cross-scene transfer while using about 3x less adaptation data. Further analyses validate the complementary gains of the leader and follower updates and reveal robust gains yet distinct learning dynamics across VLM backbones.

Junru Song, Wenhao Zhang, Yang Yang et al. · 0 citations