Projects

Other Projects

1. Computer Vision & Low-Voltage Electronics

IoT Environmental Data Collection with a Mobile Robot · 1

IoT Environmental Data Collection with a Mobile Robot

Project lead, university-level undergraduate innovation training program

Overview: Built an extensible mobile sensing platform around a Raspberry Pi 3B, integrating motion control, line following, obstacle avoidance, environmental sensors and image capture. Multiple sensors collect temperature, humidity and image data; the platform also supports 3D scene reconstruction, autonomous navigation and mobile environmental data collection.

My Role: Independently designed the system, selected hardware and developed the software. Integrated motor drivers, line-following and obstacle sensors, temperature and humidity sensors, and a vision module. Wrote low-level motion-control and sensor-acquisition code for line following, obstacle detection and movement decisions. Set up image capture and 3D scene reconstruction, assembled the vehicle, and tested the integrated system.

Outcomes: Delivered a working prototype with autonomous line following and obstacle avoidance, temperature and humidity sensing, and 3D environment reconstruction. Its modular hardware and software can accommodate additional environmental sensors and vision algorithms for mobile sensing and IoT experiments.

Design of a Linear Smoking Machine · 1

Design of a Linear Smoking Machine

2024.11 — 2025.5

Contributor to a project with China Tobacco Jiangsu Industrial Co., Ltd.

Overview: Designed and prototyped a linear smoking machine for cigarette smoke analysis. Using ISO 3308, GB/T 16450—2004 and related standards as references, designed the system around puff volume, flow rate and cycle timing. It performs smoke aspiration and collection, monitors operating conditions and records data for subsequent particulate- and gas-phase analysis.

My Role: Compared linear and rotary smoking-machine designs, then designed the aspiration, smoke collection, monitoring, control and data-acquisition modules and selected their components. Focused on the aspiration pump, filter-pad trap and cigarette holder, integrating flow and pressure sensors with PLC/STM32 control. Contributed to machining, assembly, gas-line connections and system commissioning, and optimized puff parameters and sealing.

Outcomes: Completed and commissioned an experimental prototype with adjustable puff volume, flow rate and cycle timing, plus cigarette holding, smoke collection, operating-condition monitoring and data logging. The prototype was used in the project’s smoke collection and analysis experiments and provides a platform for further equipment development and experimental work.

2. Experimental Materials Research

Processing and Property Control of High-Temperature SiC-Based Composites · 1

Processing and Property Control of High-Temperature SiC-Based Composites

2024.12 — 2025.5

Research contributor, Engineering Research Center of Advanced Lightweight Materials and Components, Ministry of Education (project budget: RMB 600,000)

Overview: Used combined slurry infiltration and polymer infiltration/pyrolysis (SI-PIP) to fabricate dense, bicontinuous Cf/(Hf,Zr,Ti)C-SiC composites for oxidizing environments across a wide temperature range. Studied matrix composition, microstructure and densification within and between fiber bundles. Related processing conditions to composition, structure, mechanical behavior, and oxidation and ablation resistance, and investigated how low-temperature pressureless densification, strengthening, toughening and oxidation/ablation protection work together.

My Role: Optimized (Hf,Zr,Ti)C slurry solids loading, dispersant dosage and pH to produce stable, low-viscosity slurries with high solids content. Deposited PyC interphases by CVI and combined vacuum/pressure-assisted slurry infiltration with SiC precursor infiltration and pyrolysis, distributing the (Hf,Zr,Ti)C-SiC matrix within and between carbon-fiber bundles while achieving low-temperature pressureless densification. Studied how SI-PIP parameters affect phase composition, matrix distribution, pore structure and density to understand densification and interfacial bonding. Measured room- and high-temperature flexural strength and fracture toughness, and examined dynamic/static oxidation and ablation over 1600–2300 ℃. Used XRD, XPS, Raman, SEM and TEM to investigate interfaces, crack propagation, fiber pullout and oxide-layer evolution, linking composition and structure to mechanical performance and oxidation/ablation resistance in Cf/(Hf,Zr,Ti)C-SiC composites.

Outcomes: 1) Validated large-component processing by fabricating 300×300 mm plates and Φ300×50 mm cylinders, and prepared 3 technical reports for project completion; 2) follow-on work received a national second prize in the Challenge Cup.

3. AI for Materials

AI for Indexing Unknown Phases in Powder X-ray Diffraction · 1

AI for Indexing Unknown Phases in Powder X-ray Diffraction

2025.06 — 2025.12

Interdisciplinary exploration driven by materials-characterization needs

Overview: Built an end-to-end CNN+Transformer multitask framework for crystal-system classification, unit-cell parameter regression and instrumental-error regression. Combined physical equations, Bragg’s law and a greedy algorithm for automated diffraction-peak matching and impurity-peak identification. Achieved AvgAcc 73.6% and RMSE 6.03 on 7014 noisy simulated XRD patterns.

My Role: Developed data preprocessing and peak detection, designed and trained the model with multitask loss weighting and physically constrained outputs, and implemented post-processing for peak matching, indexing and impurity-peak identification. Conducted error analysis and produced a reproducible algorithm workflow and technical report.

Outcomes: 1) National first prize and first place as the sole author in the 2025 7th Global Campus Artificial Intelligence Algorithm Elite Competition (Suzhou Laboratory challenge); 2) filed a Chinese invention patent application for “An integrated method for powder diffraction indexing and unit-cell refinement using physics-constrained deep learning with impurity-peak identification”, application number 202511451388.1.