Designing Smarter Systems: My EE2111A Smart Vacuum Cleaner Project

Yuxin Zhu
Apr 9
2 min read

Engineering is more than just soldering wires or simulating circuits — it’s about building functional systems grounded in logic, optimization, and real-world constraints. In EE2111A at NUS, I designed a fully theoretical but highly detailed smart robotic vacuum cleaner, featuring autonomous navigation, bin-weight detection, and AI-based energy management.

I was responsible for building the entire CAD assembly, writing the technical documentation, and guiding the group’s presentation development to meet professional standards.

Project Overview

Our vacuum robot was designed to be:

Capable of cleaning a 1000 sq ft home on one battery charge.
Able to detect bin fullness via a load cell and buzzer-triggered alert.
Auto-dock to a charging base using return-to-home logic.
Optimize energy in real time using a Physics-Informed Neural Network (PINN).
Built with low-cost, modular components, powered by STM32 or ESP32.

What I Contributed

Designed the full 3D mechanical model and component layout in Fusion 360.
Modeled internal electronics and calculated full power and runtime envelope.
Wrote the entire final design document, including all LaTeX formatting and figures.
Created high-resolution technical renders for the casing, PCB, and underside layout.
Managed team coordination and speaker notes for a cohesive presentation.

Engineering Challenges

We had to think like real product designers:

Ensure the robot could run continuously within a 36W power budget.
Calculate suction performance and cleaning coverage using physics-based models.
Select components (e.g., IRF540N, DRV8833, HX711) based on thermal, power, and cost trade-offs.
Use realistic assumptions for friction coefficients, terrain, motor speeds, and bin volume.

Project Deliverables

Want to explore our full technical work?

📎 Final Design Document (PDF) 📎 Presentation Slides (PPTX)

📎 CAD Renderings and Gallery (ZIP)

Reflections

I learned how to:

Balance hardware constraints with system-level functionality.
Translate equations into embedded control logic.
Use CAD not just for aesthetics but for manufacturability and alignment.
Write documentation that’s clean, modular, and ready for investor or faculty review.

Our final design not only meets the project specs — it represents what I’d want to prototype and build if given the chance.

Keywords: EE2111A, robotic vacuum, smart home, CAD, embedded systems, STM32, physics-informed neural networks, intelligent design, hardware optimization