Article

合成数据驱动的扩散模型用于血管腔内介入引导

基于 SOFA 的高保真手术仿真生成合成训练数据,驱动扩散模型规划血管腔内导管轨迹,克服真实数据稀缺问题。

avatar
Liu Yue
•
Synthetic-Data-Driven Diffusion Models for Endovascular Intervention Guidance featured image

Synthetic-Data-Driven Diffusion Models for Endovascular Intervention Guidance

High-fidelity SOFA-based surgical simulation generates synthetic training data for a diffusion model that plans endovascular catheter trajectories, overcoming real-data scarcity.

avatar
Liu Yue
•

面向多轴高速连续碳纤维 3D 打印的应力驱动 G-code 规划

面向多轴连续碳纤维 3D 打印的 FEA + 拓扑优化 + 测地路径规划流水线,将纤维铺放方向与主应力方向对齐以实现最优结构性能。

avatar
Liu Yue
•

Stress-Driven G-Code Planning for Multi-Axis High-Speed Continuous Carbon Fiber 3D Printing

FEA + topology optimization + geodesic path planning pipeline for multi-axis CCF 3D printing, aligning fiber deposition with principal stress directions for maximum structural …

avatar
Liu Yue
•

面向可部署视觉-语言-动作策略的高效推理基础设施

面向流匹配 VLA 策略的推理优化栈,通过投机去噪与流水线调度在商用硬件上实现 20ms 以下延迟。

avatar
Liu Yue
•

Efficient Inference Infrastructure for Deployable Vision-Language-Action Policies

Inference optimization stack for flow-matching VLA policies achieving sub-20ms latency on commodity hardware via speculative denoising and pipeline scheduling.

avatar
Liu Yue
•