Hi, I’m Longyan Pan, a master’s student in Artificial Intelligence at Shenzhen University and a member of the Guangdong Provincial Key Laboratory of Urban Spatial Information Engineering. My academic work explores spatial optimization and reinforcement learning, alongside mining large-scale trajectory data. I have also worked closely with the embodied intelligence industry, first as an algorithm intern at Yunmoxing AI and MORPHI Robot. I now work at Xianova Robotics as an Embodied Data Closed-Loop Engineer, focusing on the generalization of simulation data. My greatest aspiration is a world in which everyone can live with prosperity and freedom.
Hi，我是潘龙砚，深圳大学人工智能专业硕士研究生，隶属于广东省城市空间信息工程重点实验室。在校期间，我主要研究空间优化与强化学习，以及海量轨迹数据挖掘。我也持续走进具身智能行业，先后在陨墨星人工智能、墨奇智能担任算法实习生。目前，我在夏仙创新担任具身数据闭环工程师，专注于仿真数据泛化。我最大的愿景，是让全人类都能富裕而自由。
Advisors / 导师：Associate Professor Jizhe Xia (夏吉喆) · Assistant Professor Changxin Huang (黄畅昕)
Advisor / 导师：Dr. Yong Wu (吴勇)
具身数据闭环工程师 · 夏仙创新
Data Team / 数据组
Build a full-cycle pipeline for SparkInifinity multimodal simulation data across robot embodiments and scenarios: generation, model evaluation, and data-mix optimization. Integrate these stages into a unified feedback loop to break down silos in embodied AI data.
围绕 SparkInifinity 全模态、跨本体、跨场景仿真数据，负责数据生成、模型测评与数据配比优化，建立贯通各环节的数据闭环，打破具身智能数据孤岛。
算法实习生 · 墨奇智能
Data & Algorithms Department / 数据与算法部门
Developed multimodal whole-body human pose estimation models and built a closed-loop workflow spanning data collection, model training, evaluation, and motion optimization.
参与多模态整身人体位姿估计模型研发，搭建从数据采集、模型训练与评估到动作优化的完整闭环。
算法实习生 · 陨墨星人工智能有限公司
Algorithms Team / 算法组
Developed end-to-end and gait-perception models for exoskeleton robots. Worked across motion-capture collection, annotation, training, and on-device deployment to enable integrated perception and control.
研发外骨骼机器人端到端模型与步态感知模型，贯通动捕数据采集、标注、模型训练和板端部署，实现板端端到端感知与控制。
算法实习生 · 敦煌网
Financial Payments Department / 金融支付部门
Applied graph neural networks to model transaction patterns in B2B cross-border e-commerce and identify potentially anomalous transactions with greater precision.
运用图神经网络建模 B2B 跨境电商交易行为，提升潜在异常交易的识别精度。
Joined Xianova Robotics as an Embodied Data Closed-Loop Engineer, focusing on simulation data generalization across embodiments and scenes.
加入夏仙创新，担任具身数据闭环工程师，研究跨本体、跨场景仿真数据的泛化。
Our global structure-aware R-Tree study was published in Environmental Modelling & Software.
全局结构感知 R-Tree 研究发表于 Environmental Modelling & Software。
Joined MORPHI Robot as an algorithm intern to develop multimodal whole-body pose estimation models.
在墨奇智能担任算法实习生，参与多模态整身位姿估计模型研发。
Our hospital utilization inequalities study was published in Nature Health.
医疗服务利用不平等研究发表于 Nature Health。
Started an internship at Yunmoxing AI, developing perception and control models for exoskeleton robots.
在陨墨星人工智能担任算法实习生，研发外骨骼机器人的感知与控制模型。
Started an algorithm internship at DHgate, applying graph learning to cross-border payment risk detection.
在敦煌网开启算法实习，将图学习方法用于跨境交易异常识别。
Shijie Hou, Haotian Feng, Longyan Pan, Jizhe Xia · 2026 · DOI: 10.1016/j.envsoft.2026.107077
Jizhe Xia, Pei Ye, Longyan Pan, et al. · 2026 · DOI: 10.1038/s44360-026-00100-6
Longyan Pan, Di Zhang, Jiamei Chen, Jizhe Xia · 2026 · DOI: 10.1080/17538947.2026.2650004
Shenzhen University / 深圳大学
Shenzhen University / 深圳大学
Chongqing Normal University / 重庆师范大学
Chongqing Normal University / 重庆师范大学
Chongqing Normal University / 重庆师范大学
Department of Resource Development, Ministry of Culture and Tourism of the PRC / 中华人民共和国文化和旅游部资源开发司
Chongqing Normal University / 重庆师范大学
Chongqing Normal University / 重庆师范大学
To realize my potential amid the tides of our time.
时代浪潮中实现自我价值。