Tech Hub,
技术站是我们为数据驱动的城市研究与设计实践提供的技术引擎。我们坚信：真正有力量的设计与研究，不仅源于创意，更来自对数据与技术的深度理解。
Tech Hub serves as the technical engine that powers data-driven urban research and design practices. We believe that impactful design and inquiry are fueled not only by creativity, but also by a deep understanding of data and technology.
任务 I Mission:
我们致力于帮助设计者与研究者掌握从数据到工具的全链路操作能力。无论你希望掌握 Python + GIS + 计量 + 机器学习 + 决策优化 的空间分析流程，搭建属于自己的数据处理管线、自动化工具包或集群计算系统，还是希望深入探索 Rhino + Grasshopper + 生成式AI 的参数化建模技巧，或尝试基于 Java 的 Web 平台进行交互可视化表达，Tech Hub 都将为你提供从入门到进阶的技术支持与系统化知识体系。通过一系列模块化课程与项目实训，帮助设计者和研究者从”工具使用者“到”技术创造者“的跃迁。
We are committed to equipping designers and researchers with end-to-end technical proficiency—from data acquisition to tool deployment. Whether you're looking to master spatial analytics with Python, GIS, econometrics, machine learning, and decision optimization; build custom data processing pipelines, automation toolkits, or distributed computing systems; explore advanced parametric modeling techniques with Rhino, Grasshopper, and generative AI; or create interactive visualizations on Java-based web platforms—Tech Hub offers a comprehensive support system from foundational learning to advanced application. Through modular courses and project-based training, we empower participants to transition from tool users to technology creators.
核心组成 I Core Components:
技术站的核心板块包括四大部分：
1.全链路的空间分析技术训练：
涵盖从Python + GIS + 空间数据计量 + 机器学习在城市空间分析中的关键技术途径；
2.城市复杂系统工程训练：
致力于城市尺度的数据自动化采集与处理管线 + 深度学习 + 决策学习以及集群计算架构的搭建；
3.算法生形与生成式设计训练：
聚焦城市、建筑与景观等专业领域，探索 Rhino、Grasshopper 与生成式 AI 的协同建模方法。结合蚁群算法、羊毛毡算法、L-系统、生长算法等参数化策略，推动空间形态的规则演化与复杂结构生成。同时引入 ENVI-met 等微气候模拟工具，对建成环境进行热环境绩效评估，将环境反馈迭代入设计循环之中；通过 DepthmapX 等空间句法工具的介入，开展城市轴线分析、可视域分析、集聚度分析等空间拓扑测量，辅助设计决策中的可达性优化与场所塑造。
4.算法驱动的交互系统可视化训练：
探索基于 Java 与 Python 的 Web 平台开发，训练学员构建的交互式叙事地图、动态图表与响应式界面表达。通过前后端协同开发，实现数据、模型与空间信息的可视化联动。
the four Core Components of Research Lab are:
1) End-to-End Technical Training: Covers the essential workflows in urban spatial analysis, including Python programming, GIS, spatial econometrics, and machine learning, enabling participants to build robust analytical pipelines.
2) Urban Complex Systems Engineering: Focuses on large-scale urban data infrastructure, including automated data collection and processing pipelines, deep learning, reinforcement learning, and distributed computing architecture.
3) Building Algorithmic Generating and GenAI: Targets the domains of urbanism, architecture, and landscape architecture by exploring integrated modeling workflows using Rhino, Grasshopper, and generative AI. Participants engage with parametric strategies such as ant colony optimization, wool felt algorithms, L-systems, and growth algorithms to drive spatial form evolution and complex geometry generation.
This module also incorporates environmental simulation tools such as ENVI-met for evaluating thermal performance in built environments, enabling feedback-driven design iterations. Additionally, participants utilize tools like DepthmapX to conduct spatial topology analysis—including axial line analysis, visibility graph analysis, and clustering metrics—supporting accessibility optimization, activity hotspot identification, and spatial logic refinement in urban design.
4) Algorithm-Driven Interactive Visualization Systems: Explores the development of interactive storytelling maps, dynamic visualizations, and responsive web interfaces using Java and Python-based platforms. Through full-stack collaboration, participants learn to integrate data, models, and spatial content into cohesive, interactive visualization systems.
技术站主任 | Hub Director
佛罗里达大学，城市与区域规划博士在读
(城市灾难预测与灾后感知方向)
康奈尔大学，城市规划硕士
擅长领域：
强化学习 · 空间数据计量 · 城市韧性感知与模拟 · 智能代理建模 · 数据赋能设计 · 决策支持 · 地球遥感
研究经历:
Yu 目前在佛罗里达大学攻读城市与区域规划博士学位，研究聚焦于人工智能（AI）与强化学习算法在城市韧性响应模拟与资源动态分配的应用。他的研究致力于构建能实时感知城市状态、预测系统演化趋势并实现智能调控的城市资源配置框架，特别强调将这一前沿技术与气候韧性策略相结合。
具体而言，Yu 通过引入智能代理建模（agent-based modeling）与深度强化学习技术，研究如何在面对极端气候事件（如热浪、洪涝、干旱）时，基于历史与实时数据做出最优响应，从而提升城市系统的适应性、弹性与公平性。他的方法不仅具备高度的理论创新性，也兼具实践可操作性，能够为城市管理者提供具前瞻性的决策支持工具。
在数据赋能设计的维度上，Yu 的工作正为设计师打开一个全新视角：如何将传统设计逻辑与算法建模、预测性分析与数据驱动优化结合起来，使设计不再仅仅是静态结果，而是一个可适应、可迭代、可验证的智能系统。这种方法尤其适用于应对城市复杂问题，如土地使用冲突、应急响应配置或能源公平调配等场景。
Ph.D. Student in Urban and Regional Planning, University of Florida
(Specializing in Urban Disaster Prediction and Post-Disaster Sensing)
Master of Urban and Regional Planning, Cornell University
Expertise:
Reinforcement Learning · Spatial Data Econmetrics · Urban Resilience Simulation · Agent-based Modelling · Data-informed Design · Decision Support· Earth Remote Sensing
Research Experience:
Qianchen Yu is currently pursuing a Ph.D. in Urban and Regional Planning at the University of Florida, with research focusing on the application of artificial intelligence (AI) and reinforcement learning algorithms in urban resilience modeling and dynamic resource allocation. His work aims to develop adaptive urban systems that can sense real-time conditions, predict system evolution, and enable intelligent control—particularly when integrated with climate resilience strategies.
Specifically, Yu employs agent-based modeling and deep reinforcement learning to study how cities can make optimal, data-informed decisions in response to extreme climate events such as heatwaves, floods, and droughts. His research emphasizes both theoretical innovation and practical applicability, offering predictive, scenario-based decision support tools for urban policymakers.
In the realm of data-enabled design, Yu’s work opens up new perspectives for designers: integrating traditional design logic with algorithmic modeling, predictive analytics, and data-driven optimization. Through this approach, design evolves from a static outcome into an adaptive, iterative, and verifiable system—particularly suited to addressing complex urban challenges such as land-use conflict, emergency response planning, and equitable energy distribution.
技术站主任 | Hub Director
康奈尔大学，农业与计算机工程博士在读
(图像识别与农业智能感知方向)
康奈尔大学，电气自动化与计算机科学硕士
擅长领域：
深度学习 · 计算机视觉 · 城市图像识别 · 集群计算 · 自动化数据采集 · 决策驱动设计 · 交互式Web开发
研究经历：
Yu 目前在康奈尔大学攻读农业智能感知与计算工程博士学位，研究聚焦于人工智能在故障预测与动态控制系统中的深度应用。他的工作以数据工程与算法开发为基础，致力于构建具备感知、预测与自我调节能力的智能系统，用以提升硬件设备与城市环境中的系统响应效率与适应性。其方法论正是当前设计领域向数据驱动与智能决策转型的重要技术支撑。
在实践层面，Yu 在数据自动化采集与处理方面拥有丰富经验。他擅长利用图像识别、网络爬虫与验证码破解等技术，构建高效的信息采集流程，并借助街景图像识别技术提取建筑密度、道路形态、绿化水平与行人设施等关键空间要素，从而生成可量化的城市空间特征图谱，为设计提供数据基础与情境理解。
Yu 的加入将为团队搭建来自计算机工程与城市/建筑/景观设计交叉领域的系统性思维框架，帮助学员更深入理解“如何让数据成为设计的策略支撑”，并探索如何构建面向城市复杂性的问题识别与策略模拟，使设计与研究不仅回应问题，更具备主动预测与系统优化能力。
Ph.D. Student in Agricultural Intelligent Perception and Computational Engineering, Cornell University
(Specializing in Image Recognition and Intelligent Sensing in Agriculture)
M.S. in Electrical Automation and Computer Science, Cornell University
Expertise:
Deep Learning · Computer Vision · Urban Image Recognition · Cluster Computing · Automated Data Collection · Decision-Driven Design · Interactive Web Development.
Research Experience:
Yu is a Ph.D. candidate at Cornell University in Agricultural Intelligent Perception and Computational Engineering, specializing in AI-driven fault prediction and dynamic control systems. His research develops intelligent, self-regulating systems to enhance responsiveness in hardware and urban environments, supporting the shift toward data-driven design.
Practically, he excels in automated data acquisition, using image recognition, web scraping, and CAPTCHA bypassing to build efficient pipelines. His street-level image analysis quantifies urban features (e.g., building density, road layout, greenery) to inform design decisions.
With a computer engineering and urban design background, Yu bridges data science and design strategy, enabling AI-powered problem-solving, simulation, and optimization for smarter urban systems. His work helps teams anticipate and shape urban futures.
W
技术栈项目分享 | Tech Hub Project Showcase
Project 1 :
“Parcel Edge Classification Toolkit（PECT）：基于 Python 的场地边界分类与退界约束的可建设边界智能识别工具包开发“ | 2025Parcel Edge Classification Toolkit (PECT): A Python-based Toolkit for Parcel Edge Classification and Parcel Buildable Area Generation Based on the Zoning Setback Constraints | 2025
为解决传统场地可建设区域依赖人工标注、重复性高、效率低的问题，我们开发了 PECT ——一套结合城市用地规范的可建设区域自动生成工具包。该工具基于道路接邻关系与退界规则（如前退、后退、侧退及角地限制等），对地块边界进行智能分类，并自动生成符合当前规范的可建设边界。PECT 可广泛应用于城市设计、可开发性评估与参数化规划任务，显著提升空间设计流程中的规则嵌入能力与数据智能水平。
To address the inefficiencies of manually identifying buildable areas—often repetitive, time-consuming, and error-prone—we developed PECT, a toolkit for the automated generation of buildable zones based on urban land use regulations. Leveraging street adjacency and setback rules (such as front, rear, side, and corner constraints), PECT intelligently classifies parcel edges and generates boundary lines that comply with current zoning constraints. It is applicable to urban design, developability assessment, and parametric planning tasks, significantly enhancing the integration of regulatory logic and data intelligence within spatial design workflows.
项目二：
Project 2:
“基于 Python 的城市多中心结构自动识别与形态测度可视化工具包开发” | 2024
A Python-Based Automated Tool developing for Identifying and Measuring Urban Polycentric Morphology | 2024
为解决传统城市空间结构识别中存在的效率低、主观性强等问题，我们开发了一套基于 Python 的自动化程序，能够对城市的单中心与多中心结构进行形态测度，并将分析结果转化为可视化图像产品。该工具融合空间聚类、密度分析与多尺度指标体系，实现了从数据导入、结构识别到结果输出的全流程自动化，显著提升了城市形态研究的效率与客观性。
To address the low efficiency and subjectivity of traditional urban spatial structure analysis, we developed a Python-based automated program for measuring monocentric and polycentric urban forms, with outputs rendered as visual image products. Integrating spatial clustering, density analysis, and a multi-scale indicator system, the tool enables a fully automated workflow—from data import and structural identification to result visualization—significantly improving the efficiency and objectivity of urban morphology research.
项目三：
Project 3:
”基于 Python 的 Web 可视化平台开发：纽约曼哈顿 Uber 行驶轨迹、密度与交通拥堵的可视化“ | 2024Python-Based Web Map Platform: Visualizing Uber Trajectories, Density, and Traffic Congestion in Manhattan | 2024
为更直观理解城市出行模式与交通拥堵空间特征，我们构建了一个基于 Python 的 Web 地图叙事平台，从不同维度对纽约曼哈顿岛的 Uber 行驶数据进行可视化分析。项目聚焦于行驶轨迹、密度分布与拥堵程度等核心变量，通过轨迹数据处理、空间聚类与地图渲染，将城市交通动态转化为可交互的图像叙事，辅助分析交通热点与结构性瓶颈，并探索其与城市空间形态的关联。
To better understand urban mobility patterns and the spatial characteristics of traffic congestion, we developed a Python-based web map narrative platform that visualizes Uber trip data in Manhattan, New York. Focusing on key variables such as trip trajectories, spatial density, and congestion levels, the project integrates trajectory preprocessing, spatial clustering, and web-based map rendering to transform raw mobility data into interactive visual narratives. It supports the identification of traffic hotspots and structural bottlenecks, while revealing their spatial relationships with urban form.
项目四：
Project 4:
“基于 Mapbox GL 的 Web 地图叙事：纽约曼哈顿岛洪水风险与住房类型影响关系“ | 2024
Web Map Narrative with Mapbox GL: Visualizing the Relationship Between Flood Risk and Housing Types in Manhattan | 2024
为更直观理解纽约曼哈顿岛洪水风险等级与住房类型之间的空间关联，我们基于机器学习模型预测了 2020、2050 与 2100 年 100 年一遇洪水的空间分布，并构建了一个基于 Mapbox GL 的 Web 地图叙事平台。该平台将不同年份的洪水风险等级与住宅类型数据进行可视化叠加，揭示高风险区域与特定住房形式之间的暴露关系。通过交互式图层设计，用户可自主探索不同区域的脆弱性空间格局，为公平性导向的城市适应与气候韧性策略提供直观支持与空间参考。
To better understand the spatial relationship between flood risk levels and housing types in Manhattan, New York, we used machine learning models to predict the extent of 100-year flood events for the years 2020, 2050, and 2100. Based on these predictions, we developed a web-based map narrative platform using Mapbox GL. The platform overlays flood risk projections with housing type data, visualizing exposure patterns between high-risk areas and specific residential forms. Through interactive map layers, users can explore spatial vulnerability dynamics, providing intuitive support for equity-oriented urban adaptation and climate resilience strategies.
项目五：
Project 5:
“融合遥感影像与计算机视觉算法的非洲城市大尺度用地类型识别分析“ | 2024
Large-Scale Urban Land Cover Classification in African Cities Using Remote Sensing and Computer Vision | 2024
为填补非洲城市缺乏统一、高分辨率土地利用数据的现状，我们基于遥感影像与机器学习方法，对非洲592座城市的用地类型识别研究。项目融合多源卫星影像与分类算法，自动识别居住、工业、商业、绿地等主要城市功能区，并对其空间分布格局进行量化分析与可视化表达。该成果为区域尺度的城市规划、土地管理与资源优化配置提供了基础数据支持与空间决策依据。
To address the lack of consistent, high-resolution land use data across African cities, we conducted a large-scale land use classification study for 592 cities using remote sensing imagery and machine learning techniques. By integrating multi-source satellite data with supervised classification algorithms, the project automatically identifies major urban land use types—including residential, industrial, commercial, and green spaces—and analyzes their spatial distribution through quantitative and visual approaches. The results offer foundational data support for urban planning, land management, and resource allocation at the regional scale.
项目六：
Project 6:
“Grasshopper + GenAI 驱动的参数化设计工作流实践“ | 2024
Intelligent Parametric Design Workflow Driven by Grasshopper and Generative AI | 2024
为突破传统建模在策略生成与形式演化中存在的大量重复性操作与“不可回退”的逻辑限制，我们探索并实践了一套融合 Grasshopper 参数建模 与 生成式人工智能（GenAI） 的参数化设计工作流。该流程通过在 Grasshopper 中构建可调控的参数结构，并结合文本与图像生成工具，辅助生成初步空间原型。项目聚焦于算法生形在高级空间建模中的应用，并利用 Stable Diffusion 与 ComfyUI 作为创意参考机制，增强方案的发散性与表现力，为概念设计与竞赛策略提供高效支持。
To address the repetitive operations and irreversible logic often found in traditional modeling workflows, this project explores a design process that integrates Grasshopper-based parametric modeling with Generative AI (GenAI). The workflow constructs flexible parametric structures in Grasshopper and incorporates text-to-image tools to assist in generating early-stage spatial prototypes. It focuses on applying algorithmic form-finding to support advanced spatial modeling, while leveraging Stable Diffusion and ComfyUI as creative reference mechanisms—enhancing design divergence, visual expressiveness, and efficiency in conceptual design and competition scenarios.