Zhuang Chengbo by the San Francisco Bay during UC Berkeley Summer Sessions

Zhuang Chengbo / Mike

Zhuang Chengbo (Mike)

Future Technology (Robotics) student at Tongji University building simulation systems, digital twins, and practical AI.

I work across robotics, multi-agent simulation, semantic 3D modeling, and digital twin systems.

I care about systems that are inspectable, reproducible, maintainable, and useful beyond a classroom demo.

View Projects Get in Touch

Simulation Systems
Digital Twins & Semantic 3D
Applied AI & LLM Research

Berkeley, California · Summer 2026

Base: Shanghai, China
School: Tongji University
Track: Future Technology (Robotics)
Approach: Project Lead + Builder

Work

Selected Work

A compact view of the work that best represents my direction across digital twins, simulation systems, and applied AI.

National Undergraduate Innovation Project · Project Lead 2025 - Present

Tongji Southwest Building No.1 Digital Twin

A national-level digital twin and privacy-sensitive campus sensing project built around the iS3 platform.

Integrated multi-source campus activity data, MCP-connected AI agent services, multi-device visualization, and local encryption.
Designed MQ-135 odor sensing, LED/TCS34725 non-visual foreign-object detection, and indoor laser communication for real-time alerts in privacy-sensitive spaces.

Shanghai Municipal Innovation Project · Project Lead 2025 - Present

Generative Multimodal Semantic 3D-to-BIM System

A semantic 3D modeling workflow for smart spaces that turns visual evidence into editable, engineering-oriented BIM structure.

Uses Qwen-VL to plan observation viewpoints and SAM3 for multi-view semantic segmentation on a 3D Gaussian Splatting geometry base.
Recovers editable BIM topology through hierarchical reasoning and preserves a traceable pixel-to-Gaussian-to-mesh-to-IFC evidence chain.

UC Berkeley ENGIN 170E · Deployed Research Prototype 2026

EventShock Lab

A bilingual counterfactual market-stress laboratory for comparing matched simulations in which exactly one declared condition changes.

Combines human-reviewed evidence, frozen event packs, deterministic multi-agent simulation, and bounded LLM-assisted cognition.
Reports paired differences, uncertainty intervals, mechanism traces, and reproducibility exports without presenting synthetic paths as forecasts.

LBM Multi-Scenario Fluid Dynamics Simulator

An open-source lattice Boltzmann simulation system for multi-scenario fluid evolution, vortex dynamics, parameter studies, and visual analysis.

Covers Kármán vortex streets, complex obstacles, image mazes with BFS coupling, and NACA 0012 airfoil flow.
Supports interactive flow, vorticity, and lift visualization plus GIF/MP4 export; software copyright registered in May 2026.

Evidence

Registrations / Intellectual Property

Formal records and in-progress applications are kept separate from publications so the evidence stays accurate and easy to audit.

Software copyright registered 2026

LBM Multi-Scenario Fluid Dynamics Simulator

Software Copyright Record: 基于格子玻尔兹曼方法的多场景流体动态演化仿真系统 V1.0

Record Type: Software copyright registration
Software Name: 基于格子玻尔兹曼方法的多场景流体动态演化仿真系统
Version: V1.0
Current Status: Certificate issued
Public Identifier: Certificate No. 软著登字第17840259号; Registration No. 2026SR0625978
Key Date: Certificate date 2026-05-27
Rights: All rights; Acquisition: Original acquisition
Copyright Owners: Zhuang Chengbo / 庄程博; Ma Rujin / 马如进
Domain & Language: Computational fluid dynamics simulation; Python

Open repository

Software copyright registered 2026

Random Traffic Flow Dynamic Simulation System

Software Copyright Record: 随机车流动态仿真模拟系统 V0.2

Record Type: Software copyright registration
Software Name: 随机车流动态仿真模拟系统
Version: V0.2
Current Status: Certificate issued
Public Identifier: Certificate No. 软著登字第18081738号; Registration No. 2026SR0867457
Key Date: Certificate date 2026-08-25
Rights: All rights; Acquisition: Original acquisition
Copyright Owners: Zhuang Chengbo / 庄程博; Wang Yinyou / 王尹囿; Tang Hongzhe / 汤鸿哲; Li Zhengsheng / 李正昇; Ma Rujin / 马如进
Domain & Language: Traffic simulation; C++

Open repository

Additional Work

Other Engineering Work

Selected secondary systems that extend the main robotics, simulation, and practical AI narrative.

Robotics & multi-agent planning

Smart Warehouse AGV Path Planner

A C++17 planning core with A*, Dijkstra, and CBS multi-robot coordination, exposed through FastAPI with 2D and 3D trajectory visualization.

View repo Open live demo

Embodied AI · National research speech semifinal second prize

Mind-Matter Classroom

A Unity embodied-cognition prototype connecting MediaPipe perception and gesture input through UDP to swarm robots and shallow-water simulation.

View repo

Full-stack systems & operations

CEACStatusBot Web

A self-hosted multi-user visa-status monitoring service with scheduling, notifications, encrypted credentials, worker queues, and an admin console.

View repo

Snapshot

Education, progress, research interests, and current focus in one scan.

Instead of spreading the story across many long sections, the essentials live here as a concise snapshot.

Education

Tongji University
2025 - 2029 (Expected) | Shanghai, China
Guohao College · Future Technology Class (Robotics)
UC Berkeley Summer Sessions
Jul 6 - Aug 14, 2026 | Berkeley, CA
Completed engineering and cognitive science coursework: ENGIN 170E + COGSCI 139
Oxford Prospects and Global Development Institute
Jan 25 - Feb 7, 2026 | Oxford, UK
Short STEM programme at Regent's Park College · A+

Achievements

Shanghai Higher Education Intelligent Agent Competition (Team Track First Prize)
Shanghai Higher Education Institutions | 2026
Awarded first prize in the team track.
Annual Qiusuo Star
Tongji University | 2025 - 2026
Selected for Tongji University's annual university-level honor.
Tongji Physics Tournament (Second Prize)
Tongji University | 2025
Team award in the inaugural university physics academic competition.
Tongji Mathematical Modeling Competition (Undergraduate Third Prize)
Tongji University | 2026
Team award in the undergraduate division.

Now

Maintaining EventShock Lab as a deployed, evidence-reviewed counterfactual simulation prototype.
Developing semantic 3D-to-BIM workflows with traceable geometry and engineering outputs.
Following LLM research, especially reasoning, agents, and evaluation for practical AI systems.
Studying robotics path planning and multi-agent coordination through reproducible systems.

Contact

Reach out without hunting for the important details.

For collaboration, research discussion, or project opportunities, email is the best way to reach me.

Shanghai, China

Email Me

mike@mikezhuang.cn

CV available upon request.